Testing and bugfixing for the new MB gyp/gn wrapper.

Testing and bugfixing for the new MB gyp/gn wrapper.

This fixes a number of small issues I found while actually testing
things to get it ready to run on the bots. In particular, 'mb analyze'
didn't work right for GN. Tests were added.

TBR=brettw@chromium.org
BUG=466436
Committed: https://crrev.com/ee5b51f66b77a6d92ce46e0b119f3654deacac12
Cr-Commit-Position: refs/heads/master@{#324312}

[Dirk Pranke, 2015-04-08 21:36:59]

Patchset #1 (id:1) has been deleted

[Dirk Pranke, 2015-04-08 21:37:05]

Patchset #1 (id:20001) has been deleted

[Dirk Pranke, 2015-04-08 21:37:23]

dpranke@chromium.org changed reviewers:
+ brettw@chromium.org
- scottmg@chromium.org

[Dirk Pranke, 2015-04-08 21:37:31]

https://codereview.chromium.org/1074583002/diff/40001/tools/mb/mb.py
File tools/mb/mb.py (left):

https://codereview.chromium.org/1074583002/diff/40001/tools/mb/mb.py#oldcode288
tools/mb/mb.py:288: raise MBErr('Cycle in mixins for "%s": %s' % (m, visited))
This logic didn't actually work right; it would err out if the config repeated a
mixin, even if it wasn't a true cycle. 

Arguably, repeated mixins should also be bad, but I'm still on the fence about
that.

[Dirk Pranke, 2015-04-08 21:37:34]

The CQ bit was checked by dpranke@chromium.org

[commit-bot: I haz the power, 2015-04-08 21:37:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1074583002/40001

[commit-bot: I haz the power, 2015-04-09 00:03:29]

Committed patchset #1 (id:40001)

[commit-bot: I haz the power, 2015-04-09 00:04:19]

Patchset 1 (id:??) landed as
https://crrev.com/ee5b51f66b77a6d92ce46e0b119f3654deacac12
Cr-Commit-Position: refs/heads/master@{#324312}