Add test for canonical host alternative service.

Add test for canonical host alternative service.

If a canonical server advertises an alternative service without an explicit
hostname, or an alternative protocol, then it is implied that the same
alternative service works on other hosts.  This CL adds a test to verify that if
SetAlternativeService is called with an empty host (denoting no explicit
hostname in the alternative service advertisement), then
GetAlternativeService(origin) returns alternative services with host ==
origin.host().  This will be important once HttpServerProperties starts storing
alternative services with hostname to make sure current behavior is preserved.

BUG=392576
Committed: https://crrev.com/0186dcd1238a84cfe71134ee24e3602ff0404af6
Cr-Commit-Position: refs/heads/master@{#321976}

[Bence, 2015-03-23 17:54:10]

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

[Bence, 2015-03-23 17:54:10]

Ryan, PTAL.  Thanks.

[Ryan Hamilton, 2015-03-23 20:48:07]

lgtm

https://codereview.chromium.org/1029633002/diff/1/net/http/http_server_proper...
File net/http/http_server_properties_impl_unittest.cc (right):

https://codereview.chromium.org/1029633002/diff/1/net/http/http_server_proper...
net/http/http_server_properties_impl_unittest.cc:461:
EXPECT_EQ(test_host_port_pair.host(), alternative_service.host);
This is great! Nice.

That being said, I'm astounded that it works! :> Did you explicitly implement
this behavior in an earlier CL?

[Bence, 2015-03-24 12:03:38]

The CQ bit was checked by bnc@chromium.org

[commit-bot: I haz the power, 2015-03-24 12:03:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1029633002/1

[Bence, 2015-03-24 12:05:34]

On 2015/03/23 20:48:07, Ryan Hamilton wrote:
> lgtm
> 
>
https://codereview.chromium.org/1029633002/diff/1/net/http/http_server_proper...
> File net/http/http_server_properties_impl_unittest.cc (right):
> 
>
https://codereview.chromium.org/1029633002/diff/1/net/http/http_server_proper...
> net/http/http_server_properties_impl_unittest.cc:461:
> EXPECT_EQ(test_host_port_pair.host(), alternative_service.host);
> This is great! Nice.
> 
> That being said, I'm astounded that it works! :> Did you explicitly implement
> this behavior in an earlier CL?

We inherited this for free from the alternate protocol world.  First,
GetAlternateProtocol only returned a (port, protocol) pair.  When I made it
return an AlternativeService, I injected the hostname of origin (not the
canonical host) to preserve behavior.  Now this test is here to document this,
so that when we store AlternativeServiceInfo instead of AlternateProtocolInfo
for origins, we'll be forced to pay attention to this.

[commit-bot: I haz the power, 2015-03-24 12:11:25]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-03-24 12:12:19]

Patchset 1 (id:??) landed as
https://crrev.com/0186dcd1238a84cfe71134ee24e3602ff0404af6
Cr-Commit-Position: refs/heads/master@{#321976}