Revert of Remove EventSender from ImageLoader

Revert of Remove EventSender from ImageLoader (patchset #10 id:180001 of https://codereview.chromium.org/2863763003/ )

Reason for revert:
This patch probably causes object leaks:

See details in "webkit_tests" of
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty%20Leak/builds/4925


Original issue's description:
> Remove EventSender from ImageLoader
> 
> Instead of using EventSender, this CL schedules ImageLoader's
> load and error events using PostCancellableTask.
> has_pending_load_event_ and has_pending_error_event_ booleans are
> replaced with TaskHandle to the cancellable tasks and IsActive() calls.
> 
> This CL removes the dependencies between Document's load event and
> <img> load events via EventSender.
> 
> This CL removes EventSender.h as ImageLoader was the last user of that.
> 
> Test changes:
> stop-loading.html:
>   testharness.js waits for the window load event before finishing.
>   Previously, we call window.stop() in <img> load event, but window
>   load event is fired, because the <img> load event is fired from
>   Document::ImplicitClose() via EventSender. At that time window.stop()
>   doesn't stop the window load event.
>   After this CL, <img> load event is fired separately from
>   Document::ImplicitClose(), and thus window.stop() stops the window
>   load event.
>   To work around this, this CL triggers window load event explicitly
>   just to signal testharness.js to finish.
> 
> http/tests/loading/preload-image-sizes-2x.html and
> http/tests/loading/preload-picture-sizes-2x.html:
>   The tests uses srcset-helper.js that reloads the page with empty
>   cache in the window load event.
>   Because this CL changes the timing of the window load event, this
>   causes some asynchronous tasks that cause new image loading to be
>   scheduled before window load event and executed after that, causing
>   some images left on the cache before reloading starts.
>   This CL works around the issue by waiting a little bit for those
>   tasks to be executed, and then clearing the cache and reloading the
>   page.
> 
> BUG=624697, 720268
> 
> Review-Url: https://codereview.chromium.org/2863763003
> Cr-Commit-Position: refs/heads/master@{#474086}
> Committed: https://chromium.googlesource.com/chromium/src/+/da70e32d811fffe4e15728ba858dbdfa3180ed59

TBR=japhet@chromium.org,yhirano@chromium.org,kinuko@chromium.org,kouhei@chromium.org,tzik@chromium.org,hiroshige@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=624697, 720268
Review-Url: https://codereview.chromium.org/2903773002
Cr-Commit-Position: refs/heads/master@{#474154}
Committed: https://chromium.googlesource.com/chromium/src/+/8e7dc7edbcacdd495eb18ff33fe88d1fb89a15d5

[yosin_UTC9, 2017-05-24 04:06:13]

The CQ bit was checked by yosin@chromium.org

[yosin_UTC9, 2017-05-24 04:06:13]

Created Revert of Remove EventSender from ImageLoader

[commit-bot: I haz the power, 2017-05-24 04:07:00]

CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-05-24 04:07:32]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1495598773212590, "parent_rev":
"cca67d31f4a0b4ec2c20090a8bfc38c464274ab2", "commit_rev":
"8e7dc7edbcacdd495eb18ff33fe88d1fb89a15d5"}

[commit-bot: I haz the power, 2017-05-24 04:07:48]

Description was changed from

==========
Revert of Remove EventSender from ImageLoader (patchset #10 id:180001 of
https://codereview.chromium.org/2863763003/ )

Reason for revert:
This patch probably causes object leaks:

See details in "webkit_tests" of
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty...


Original issue's description:
> Remove EventSender from ImageLoader
> 
> Instead of using EventSender, this CL schedules ImageLoader's
> load and error events using PostCancellableTask.
> has_pending_load_event_ and has_pending_error_event_ booleans are
> replaced with TaskHandle to the cancellable tasks and IsActive() calls.
> 
> This CL removes the dependencies between Document's load event and
> <img> load events via EventSender.
> 
> This CL removes EventSender.h as ImageLoader was the last user of that.
> 
> Test changes:
> stop-loading.html:
>   testharness.js waits for the window load event before finishing.
>   Previously, we call window.stop() in <img> load event, but window
>   load event is fired, because the <img> load event is fired from
>   Document::ImplicitClose() via EventSender. At that time window.stop()
>   doesn't stop the window load event.
>   After this CL, <img> load event is fired separately from
>   Document::ImplicitClose(), and thus window.stop() stops the window
>   load event.
>   To work around this, this CL triggers window load event explicitly
>   just to signal testharness.js to finish.
> 
> http/tests/loading/preload-image-sizes-2x.html and
> http/tests/loading/preload-picture-sizes-2x.html:
>   The tests uses srcset-helper.js that reloads the page with empty
>   cache in the window load event.
>   Because this CL changes the timing of the window load event, this
>   causes some asynchronous tasks that cause new image loading to be
>   scheduled before window load event and executed after that, causing
>   some images left on the cache before reloading starts.
>   This CL works around the issue by waiting a little bit for those
>   tasks to be executed, and then clearing the cache and reloading the
>   page.
> 
> BUG=624697, 720268
> 
> Review-Url: https://codereview.chromium.org/2863763003
> Cr-Commit-Position: refs/heads/master@{#474086}
> Committed:
https://chromium.googlesource.com/chromium/src/+/da70e32d811fffe4e15728ba858d...

TBR=japhet@chromium.org,yhirano@chromium.org,kinuko@chromium.org,kouhei@chrom...
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=624697, 720268
==========

to

==========
Revert of Remove EventSender from ImageLoader (patchset #10 id:180001 of
https://codereview.chromium.org/2863763003/ )

Reason for revert:
This patch probably causes object leaks:

See details in "webkit_tests" of
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty...


Original issue's description:
> Remove EventSender from ImageLoader
> 
> Instead of using EventSender, this CL schedules ImageLoader's
> load and error events using PostCancellableTask.
> has_pending_load_event_ and has_pending_error_event_ booleans are
> replaced with TaskHandle to the cancellable tasks and IsActive() calls.
> 
> This CL removes the dependencies between Document's load event and
> <img> load events via EventSender.
> 
> This CL removes EventSender.h as ImageLoader was the last user of that.
> 
> Test changes:
> stop-loading.html:
>   testharness.js waits for the window load event before finishing.
>   Previously, we call window.stop() in <img> load event, but window
>   load event is fired, because the <img> load event is fired from
>   Document::ImplicitClose() via EventSender. At that time window.stop()
>   doesn't stop the window load event.
>   After this CL, <img> load event is fired separately from
>   Document::ImplicitClose(), and thus window.stop() stops the window
>   load event.
>   To work around this, this CL triggers window load event explicitly
>   just to signal testharness.js to finish.
> 
> http/tests/loading/preload-image-sizes-2x.html and
> http/tests/loading/preload-picture-sizes-2x.html:
>   The tests uses srcset-helper.js that reloads the page with empty
>   cache in the window load event.
>   Because this CL changes the timing of the window load event, this
>   causes some asynchronous tasks that cause new image loading to be
>   scheduled before window load event and executed after that, causing
>   some images left on the cache before reloading starts.
>   This CL works around the issue by waiting a little bit for those
>   tasks to be executed, and then clearing the cache and reloading the
>   page.
> 
> BUG=624697, 720268
> 
> Review-Url: https://codereview.chromium.org/2863763003
> Cr-Commit-Position: refs/heads/master@{#474086}
> Committed:
https://chromium.googlesource.com/chromium/src/+/da70e32d811fffe4e15728ba858d...

TBR=japhet@chromium.org,yhirano@chromium.org,kinuko@chromium.org,kouhei@chrom...
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=624697, 720268

Review-Url: https://codereview.chromium.org/2903773002
Cr-Commit-Position: refs/heads/master@{#474154}
Committed:
https://chromium.googlesource.com/chromium/src/+/8e7dc7edbcacdd495eb18ff33fe8...
==========

[commit-bot: I haz the power, 2017-05-24 04:07:49]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/chromium/src/+/8e7dc7edbcacdd495eb18ff33fe8...

[hiroshige, 2017-05-26 21:23:06]

lgtm

[hiroshige, 2017-05-26 21:24:27]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/2906063003/ by hiroshige@chromium.org.

The reason for reverting is: Trying reland with fixes..