[SPDY] Use WeakPtr<SpdySession> everywhere but SpdySessionPool

[SPDY] Use WeakPtr<SpdySession> everywhere but SpdySessionPool

Make SpdyHttpStream cache it's SPDY stream's LoadTimingInfo
on close. Also, make SpdyHttpStream query SpdySession::IsReused()
when it's constructed and cache the value, as that's closer to the
intent of its use.

Avoid uses of SpdySession::IsClosed() in non-test code and add
TODO to remove uses from test code. This is more correct since
testing the WeakPtr is a stronger condition than testing openness
and SpdySession functions already do the right thing if the
SpdySession is already closed.

Tweak some tests that implicitly depended on
having refs to SpdySession.

BUG=255701
R=rch@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=212858

[akalin, 2013-07-13 02:22:05]

+rch for review

[Ryan Hamilton, 2013-07-18 15:33:28]

lgtm!  I'm not sure if we'll want to expose WeakPtr<SpdySession> publicly, but
this is an awesome step towards non-refcounting of SpdySessions.

[akalin, 2013-07-22 07:04:49]

Fixed a test (newly added by the previous CL), did some minor formatting and
comment changes, and also replaced some SpdySession::IsClosed() tests with
testing the WeakPtr itself in non-test code (and added a TODO for test code).
This is more correct since testing the WeakPtr is a stronger condition than open
vs. closed, and SpdySession functions already do the right thing if the
SpdySession is already closed.

[akalin, 2013-07-22 07:05:45]

On 2013/07/22 07:04:49, akalin wrote:
> Fixed a test (newly added by the previous CL), did some minor formatting and
> comment changes, and also replaced some SpdySession::IsClosed() tests with
> testing the WeakPtr itself in non-test code (and added a TODO for test code).
> This is more correct since testing the WeakPtr is a stronger condition than
open
> vs. closed, and SpdySession functions already do the right thing if the
> SpdySession is already closed.

Committing soon!

[akalin, 2013-07-22 09:37:40]

Committed patchset #4 manually as r212858 (presubmit successful).

[akalin, 2013-07-22 09:44:27]

On 2013/07/22 09:37:40, akalin wrote:
> Committed patchset #4 manually as r212858 (presubmit successful).

Ugh, I must have been looking at the wrong CL page -- I thought the trybots were
green.

I guess I'll keep an eye and will revert if it breaks the tree.