Fix potential bug in push_test.js - do not cache subscription options

Fix potential bug in push_test.js - do not cache subscription options

- Prior to this patch, the application server key could be inadvertently
  cached, e.g. if a test in PushMessagingBrowserTest did the following:
    documentSubscribePush()
    removeManifest()
    documentSubscribePushWithoutKey()
  then the second push would actually use the cached key from the first.

- No existing tests happen to hit the bug but that's just luck.

BUG=660874
Review-Url: https://codereview.chromium.org/2460223004
Cr-Commit-Position: refs/heads/master@{#428983}
(cherry picked from commit 5069448246d980f994a700bc937aba59dc6e9ab6)

Rebased to adjust documentSubscribePushBadKey (which had been removed by
https://codereview.chromium.org/2411733002 on master).

Merge approval: https://crbug.com/659230#c25

Committed: https://chromium.googlesource.com/chromium/src/+/59705097c9ca9ee36d441860aaf06e00a70b8bff

[johnme, 2016-11-21 14:43:43]

Description was changed from

==========
Fix potential bug in push_test.js - do not cache subscription options

- Prior to this patch, the application server key could be inadvertently
  cached, e.g. if a test in PushMessagingBrowserTest did the following:
    documentSubscribePush()
    removeManifest()
    documentSubscribePushWithoutKey()
  then the second push would actually use the cached key from the first.

- No existing tests happen to hit the bug but that's just luck.

BUG=660874

Review-Url: https://codereview.chromium.org/2460223004
Cr-Commit-Position: refs/heads/master@{#428983}
(cherry picked from commit 5069448246d980f994a700bc937aba59dc6e9ab6)

Rebased to adjust documentSubscribePushBadKey (which had been removed by
https://codereview.chromium.org/2411733002 on master).

Merge approval: https://crbug.com/659230#c25
==========

to

==========
Fix potential bug in push_test.js - do not cache subscription options

- Prior to this patch, the application server key could be inadvertently
  cached, e.g. if a test in PushMessagingBrowserTest did the following:
    documentSubscribePush()
    removeManifest()
    documentSubscribePushWithoutKey()
  then the second push would actually use the cached key from the first.

- No existing tests happen to hit the bug but that's just luck.

BUG=660874

Review-Url: https://codereview.chromium.org/2460223004
Cr-Commit-Position: refs/heads/master@{#428983}
(cherry picked from commit 5069448246d980f994a700bc937aba59dc6e9ab6)

Rebased to adjust documentSubscribePushBadKey (which had been removed by
https://codereview.chromium.org/2411733002 on master).

Merge approval: https://crbug.com/659230#c25

Committed:
https://chromium.googlesource.com/chromium/src/+/59705097c9ca9ee36d441860aaf0...
==========

[johnme, 2016-11-21 14:43:45]

Committed patchset #1 (id:1) manually as
59705097c9ca9ee36d441860aaf06e00a70b8bff.