Handle recovery on "timestamp-looking" lines in the WebVTT parser

Handle recovery on "timestamp-looking" lines in the WebVTT parser

When a line with a "-->" is encountered as part of cue-text, in the
header part or after a previous error (i.e. during "bad cue" processing).
The parser is supposed to break out of the current state and try to parse
the "-->" line as being the start of a new cue.

Updated testcases:

PASS -> FAIL
tc006, tc007 and tc011 (Tested the old behavior where "-->" was
allowed/ignored, and blank lines were required to start a new cue.)

FAIL -> PASS
opera/track/webvtt/parsing subtests nulls, header-timings, header-space,
header-tab and arrows
opera/interfaces/TextTrackCue/vertical.html

TEST=media/track/track-webvtt-tc031-cue-recovery.html
TEST=media/track/track-webvtt-tc032-degenerate-cues.html
BUG=305317
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=162210

[jochen (gone - plz use gerrit), 2013-11-18 13:28:40]

lgtm

I wonder whether it's worthwhile to introduce unit tests for VTTParser?

[fs, 2013-11-18 13:41:25]

On 2013/11/18 13:28:40, jochen wrote:
> lgtm
> 
> I wonder whether it's worthwhile to introduce unit tests for VTTParser?

For some of the parts it should probably be possible (all the static helpers).
Any wider scope would probably require refactoring to get rid of the Document
reference (or at least no require it for cue emittal.)

[commit-bot: I haz the power, 2013-11-18 13:41:59]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fs@opera.com/75273002/1

[commit-bot: I haz the power, 2013-11-18 16:09:25]

Change committed as 162210