Revert of Replace FFmpegDemuxer thread per element with base::TaskScheduler.

Revert of Replace FFmpegDemuxer thread per element with base::TaskScheduler. (patchset #8 id:140001 of https://codereview.chromium.org/2710133003/ )

Reason for revert:
It's most likely a culprit for failing test:
http/tests/media/video-in-iframe-crash.html starting from
https://uberchromegw.corp.google.com/i/chromium.webkit/builders/WebKit%20Linux%20Trusty/builds/23846

Error message:
[ERROR:render_media_log.cc(30)] MediaEvent: MEDIA_ERROR_LOG_ENTRY {"error":"FFmpegDemuxer: data source error"}

Original issue's description:
> Replace FFmpegDemuxer thread per element with base::TaskScheduler.
> 
> I don't think we need to keep starting a thread per <audio> or
> <video> element anymore. Instead we can use a shared pool that
> only spins up a new thread when needed.
> 
> This change switches FFmpegDemuxer from creating its own thread
> to using base::TaskScheduler for executing blocking network reads.
> The TaskScheduler API is perfectly suited for our use case and
> required little changes.
> 
> Doing this required some changes to our BlockingUrlProtocol to
> make it safe to access after FFmpegDemuxer and potentially the
> DataSource have been destroyed.
> 
> This should reduce the number of out of threads crashes we see on
> a fairly regular basis.
> 
> BUG=61293
> TEST=existing tests all pass, no slowdown noticed on pages like gfycat.com
> 
> Review-Url: https://codereview.chromium.org/2710133003
> Cr-Commit-Position: refs/heads/master@{#453470}
> Committed: https://chromium.googlesource.com/chromium/src/+/5384b9fe1cf67869e1bf98ab8ff6bd21ae4216da

TBR=gab@chromium.org,halliwell@chromium.org,xhwang@chromium.org,dalecurtis@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=61293
Review-Url: https://codereview.chromium.org/2719883005
Cr-Commit-Position: refs/heads/master@{#453563}
Committed: https://chromium.googlesource.com/chromium/src/+/584b59ac597f611fbac882fbbe90d538bf60db4e

[dvadym, 2017-02-28 10:13:19]

The CQ bit was checked by dvadym@chromium.org

[dvadym, 2017-02-28 10:13:19]

Created Revert of Replace FFmpegDemuxer thread per element with
base::TaskScheduler.

[commit-bot: I haz the power, 2017-02-28 10:13:27]

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, 2017-02-28 10:14:10]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1488276799007180, "parent_rev":
"f9ae27fa739d2a97c7c9d546d71b1aac8dcc442e", "commit_rev":
"584b59ac597f611fbac882fbbe90d538bf60db4e"}

[commit-bot: I haz the power, 2017-02-28 10:14:44]

Description was changed from

==========
Revert of Replace FFmpegDemuxer thread per element with base::TaskScheduler.
(patchset #8 id:140001 of https://codereview.chromium.org/2710133003/ )

Reason for revert:
It's most likely a culprit for failing test:
http/tests/media/video-in-iframe-crash.html starting from
https://uberchromegw.corp.google.com/i/chromium.webkit/builders/WebKit%20Linu...

Error message:
[ERROR:render_media_log.cc(30)] MediaEvent: MEDIA_ERROR_LOG_ENTRY
{"error":"FFmpegDemuxer: data source error"}

Original issue's description:
> Replace FFmpegDemuxer thread per element with base::TaskScheduler.
> 
> I don't think we need to keep starting a thread per <audio> or
> <video> element anymore. Instead we can use a shared pool that
> only spins up a new thread when needed.
> 
> This change switches FFmpegDemuxer from creating its own thread
> to using base::TaskScheduler for executing blocking network reads.
> The TaskScheduler API is perfectly suited for our use case and
> required little changes.
> 
> Doing this required some changes to our BlockingUrlProtocol to
> make it safe to access after FFmpegDemuxer and potentially the
> DataSource have been destroyed.
> 
> This should reduce the number of out of threads crashes we see on
> a fairly regular basis.
> 
> BUG=61293
> TEST=existing tests all pass, no slowdown noticed on pages like gfycat.com
> 
> Review-Url: https://codereview.chromium.org/2710133003
> Cr-Commit-Position: refs/heads/master@{#453470}
> Committed:
https://chromium.googlesource.com/chromium/src/+/5384b9fe1cf67869e1bf98ab8ff6...

TBR=gab@chromium.org,halliwell@chromium.org,xhwang@chromium.org,dalecurtis@ch...
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=61293
==========

to

==========
Revert of Replace FFmpegDemuxer thread per element with base::TaskScheduler.
(patchset #8 id:140001 of https://codereview.chromium.org/2710133003/ )

Reason for revert:
It's most likely a culprit for failing test:
http/tests/media/video-in-iframe-crash.html starting from
https://uberchromegw.corp.google.com/i/chromium.webkit/builders/WebKit%20Linu...

Error message:
[ERROR:render_media_log.cc(30)] MediaEvent: MEDIA_ERROR_LOG_ENTRY
{"error":"FFmpegDemuxer: data source error"}

Original issue's description:
> Replace FFmpegDemuxer thread per element with base::TaskScheduler.
> 
> I don't think we need to keep starting a thread per <audio> or
> <video> element anymore. Instead we can use a shared pool that
> only spins up a new thread when needed.
> 
> This change switches FFmpegDemuxer from creating its own thread
> to using base::TaskScheduler for executing blocking network reads.
> The TaskScheduler API is perfectly suited for our use case and
> required little changes.
> 
> Doing this required some changes to our BlockingUrlProtocol to
> make it safe to access after FFmpegDemuxer and potentially the
> DataSource have been destroyed.
> 
> This should reduce the number of out of threads crashes we see on
> a fairly regular basis.
> 
> BUG=61293
> TEST=existing tests all pass, no slowdown noticed on pages like gfycat.com
> 
> Review-Url: https://codereview.chromium.org/2710133003
> Cr-Commit-Position: refs/heads/master@{#453470}
> Committed:
https://chromium.googlesource.com/chromium/src/+/5384b9fe1cf67869e1bf98ab8ff6...

TBR=gab@chromium.org,halliwell@chromium.org,xhwang@chromium.org,dalecurtis@ch...
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=61293

Review-Url: https://codereview.chromium.org/2719883005
Cr-Commit-Position: refs/heads/master@{#453563}
Committed:
https://chromium.googlesource.com/chromium/src/+/584b59ac597f611fbac882fbbe90...
==========

[commit-bot: I haz the power, 2017-02-28 10:14:46]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/chromium/src/+/584b59ac597f611fbac882fbbe90...