Revert 203489 "Replace JSON (de)serialization of extension messa..."

Revert 203489 "Replace JSON (de)serialization of extension messa..."

Breaks page_cycler_2012Q2-netsim benchmark.

BUG=246213
> Replace JSON (de)serialization of extension messages with direct Value pickling.
> 
> BUG=55316
> R=mpcomplete@chromium.org
> TBR=cdn@chromium.org
> 
> Review URL: https://chromiumcodereview.appspot.com/16226004

TBR=kalman@chromium.org
NOTRY=True

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=203753

[tonyg, 2013-06-03 19:05:01]

Seems like it is time to get the tree green again and continue investigating
this breakage off of the tree. Could you please lgtm this revert?

[not at google - send to devlin, 2013-06-03 19:07:00]

damn it. the problem is actually https://chromiumcodereview.appspot.com/10837066

a quick fix without reverting this patch would be to change the perf tests to
save a reference to chrome.performance.timing before sending it.

lgtm, but it would save me a lot of time today if you would be able to do that
instead.

[tonyg, 2013-06-03 19:12:28]

On 2013/06/03 19:07:00, kalman wrote:
> damn it. the problem is actually
https://chromiumcodereview.appspot.com/10837066

Is that the right patch? It is 10 months old and this failure just started with
this patch and it is fixed with reverting this patch.

> 
> a quick fix without reverting this patch would be to change the perf tests to
> save a reference to chrome.performance.timing before sending it.
> 

I don't understand. Even if there is a workaround on the extension side, we
can't break existing extensions, can we?

> lgtm, but it would save me a lot of time today if you would be able to do that
> instead.

I'm going to proceed with the rollback. Rolling forward with the fix should be
equally quick. I'm just trying to sheriff the waterfall and don't have the time
to continue investigating this breakage.

[tonyg, 2013-06-03 19:26:33]

Committed patchset #1 manually as r203753 (presubmit successful).