Fix dangling reference in NormalSpdyTransactionHelper.

Fix dangling reference in NormalSpdyTransactionHelper.

NormalSpdyTransactionHelper constructor is often (approximately 70 times) called
with a temporary BoundNetLog() argument.  Its log_ member currently takes
reference of that, which becomes dangling right after the constructor returns.

This is not an issue if there are no NetLog events, nor in the NetLog test which
creates a named CapturingBoundNetLog object that it keeps in scope.  The
dangling reference only became problematic in the HTTP11RequiredProxyRetry test,
and curiously enough, only for 64 bit Andoid builds.

This CL changes the NormalSpdyTransactionHelper.log_ member type from reference
to value to fix this issue.

BUG=447837
Committed: https://crrev.com/29b1f07f9291ca0358117efb691cae9ae6ded989
Cr-Commit-Position: refs/heads/master@{#311538}

[Bence, 2015-01-14 18:53:49]

bnc@chromium.org changed reviewers:
+ mmenke@chromium.org

[Bence, 2015-01-14 18:53:50]

mmenke: PTAL.  Thank you.

[mmenke, 2015-01-14 18:54:24]

On 2015/01/14 18:53:50, Bence wrote:
> mmenke: PTAL.  Thank you.

LGTM

[Bence, 2015-01-14 20:33:02]

The CQ bit was checked by bnc@chromium.org

[commit-bot: I haz the power, 2015-01-14 20:33:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/852053002/1

[commit-bot: I haz the power, 2015-01-14 20:38:18]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-01-14 20:39:39]

Patchset 1 (id:??) landed as
https://crrev.com/29b1f07f9291ca0358117efb691cae9ae6ded989
Cr-Commit-Position: refs/heads/master@{#311538}

[Ryan Hamilton, 2015-01-15 20:31:41]

On 2015/01/14 18:54:24, mmenke wrote:
> On 2015/01/14 18:53:50, Bence wrote:
> > mmenke: PTAL.  Thank you.
> 
> LGTM

Nice catch! Do you think this was the same issue we saw with the other
HTTP_1_1_REQUIRED test? If so, is it worth trying that tests again?

[Bence, 2015-01-15 20:49:55]

On 2015/01/15 20:31:41, Ryan Hamilton wrote:
> On 2015/01/14 18:54:24, mmenke wrote:
> > On 2015/01/14 18:53:50, Bence wrote:
> > > mmenke: PTAL.  Thank you.
> > 
> > LGTM
> 
> Nice catch! Do you think this was the same issue we saw with the other
> HTTP_1_1_REQUIRED test? If so, is it worth trying that tests again?

Good call, I'll try.