Revert of Don't use pending NavigationEntries for navigation transfers.

Revert of Don't use pending NavigationEntries for navigation transfers. (patchset #5 id:80001 of https://codereview.chromium.org/1683593002/ )

Reason for revert:
This is causing crashes in https://crbug.com/600743, as well as a problem with interstitial pages in https://crbug.com/600046.  I'll look into the problems and then reland this.

Original issue's description:
> Don't use pending NavigationEntries for navigation transfers.
>
> Cross-process transfers for navigations that are about to commit
> had been using NavigationControllerImpl::LoadURLWithParams (as
> an artifact of previously going through OpenURL).  Most of that
> code is unnecessary, in particular the fact that it creates a
> new pending NavigationEntry.  That's problematic for subframe
> transfers, where we should not affect any existing pending
> NavigationEntry (e.g., for a slow main frame navigation).
>
> This CL shortcuts the transfer by creating a NavigationEntry
> without making it the pending one.  A future CL can eliminate
> the entry entirely by validating and passing the parameters
> to the right RenderFrameHost directly.
>
> This also makes it possible to do cross-process navigations
> in subframes of the initial blank page, fixing a crash.
>
> BUG=495161, 584739
> TEST=Create OOPIF on initial blank page of a tab.
> TEST=Cross-process subframe navigation during slow main frame navigation.
> CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation
>
> Committed: https://crrev.com/9bc97a7b5405403cd9878556162a6234ef4381bf
> Cr-Commit-Position: refs/heads/master@{#384009}

TBR=alexmos@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=495161, 584739
Committed: https://crrev.com/e737670f4917aa311ec5c583d9e1ac9386d46225
Cr-Commit-Position: refs/heads/master@{#385235}

[Charlie Reis, 2016-04-05 17:28:59]

Created Revert of Don't use pending NavigationEntries for navigation transfers.

[commit-bot: I haz the power, 2016-04-05 17:29:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1861773002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1861773002/1

[commit-bot: I haz the power, 2016-04-05 17:31:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1861773002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1861773002/1

[commit-bot: I haz the power, 2016-04-05 18:01:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1861773002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1861773002/1

[commit-bot: I haz the power, 2016-04-05 18:31:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1861773002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1861773002/1

[commit-bot: I haz the power, 2016-04-05 18:38:10]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-04-05 18:40:20]

Description was changed from

==========
Revert of Don't use pending NavigationEntries for navigation transfers.
(patchset #5 id:80001 of https://codereview.chromium.org/1683593002/ )

Reason for revert:
This is causing crashes in https://crbug.com/600743, as well as a problem with
interstitial pages in https://crbug.com/600046.  I'll look into the problems and
then reland this.

Original issue's description:
> Don't use pending NavigationEntries for navigation transfers.
> 
> Cross-process transfers for navigations that are about to commit
> had been using NavigationControllerImpl::LoadURLWithParams (as
> an artifact of previously going through OpenURL).  Most of that
> code is unnecessary, in particular the fact that it creates a
> new pending NavigationEntry.  That's problematic for subframe
> transfers, where we should not affect any existing pending
> NavigationEntry (e.g., for a slow main frame navigation).
> 
> This CL shortcuts the transfer by creating a NavigationEntry
> without making it the pending one.  A future CL can eliminate
> the entry entirely by validating and passing the parameters
> to the right RenderFrameHost directly.
> 
> This also makes it possible to do cross-process navigations
> in subframes of the initial blank page, fixing a crash.
> 
> BUG=495161, 584739
> TEST=Create OOPIF on initial blank page of a tab.
> TEST=Cross-process subframe navigation during slow main frame navigation.
> CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation
> 
> Committed: https://crrev.com/9bc97a7b5405403cd9878556162a6234ef4381bf
> Cr-Commit-Position: refs/heads/master@{#384009}

TBR=alexmos@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=495161, 584739
==========

to

==========
Revert of Don't use pending NavigationEntries for navigation transfers.
(patchset #5 id:80001 of https://codereview.chromium.org/1683593002/ )

Reason for revert:
This is causing crashes in https://crbug.com/600743, as well as a problem with
interstitial pages in https://crbug.com/600046.  I'll look into the problems and
then reland this.

Original issue's description:
> Don't use pending NavigationEntries for navigation transfers.
>
> Cross-process transfers for navigations that are about to commit
> had been using NavigationControllerImpl::LoadURLWithParams (as
> an artifact of previously going through OpenURL).  Most of that
> code is unnecessary, in particular the fact that it creates a
> new pending NavigationEntry.  That's problematic for subframe
> transfers, where we should not affect any existing pending
> NavigationEntry (e.g., for a slow main frame navigation).
>
> This CL shortcuts the transfer by creating a NavigationEntry
> without making it the pending one.  A future CL can eliminate
> the entry entirely by validating and passing the parameters
> to the right RenderFrameHost directly.
>
> This also makes it possible to do cross-process navigations
> in subframes of the initial blank page, fixing a crash.
>
> BUG=495161, 584739
> TEST=Create OOPIF on initial blank page of a tab.
> TEST=Cross-process subframe navigation during slow main frame navigation.
> CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation
>
> Committed: https://crrev.com/9bc97a7b5405403cd9878556162a6234ef4381bf
> Cr-Commit-Position: refs/heads/master@{#384009}

TBR=alexmos@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=495161, 584739

Committed: https://crrev.com/e737670f4917aa311ec5c583d9e1ac9386d46225
Cr-Commit-Position: refs/heads/master@{#385235}
==========

[commit-bot: I haz the power, 2016-04-05 18:40:21]

Patchset 1 (id:??) landed as
https://crrev.com/e737670f4917aa311ec5c583d9e1ac9386d46225
Cr-Commit-Position: refs/heads/master@{#385235}