Remove temporary hack from checks_instance_lock.

The checks_instance_lock decorator was temporarily made a mess of while
the compute manager was being converted over to taking an instance
instead of instance_uuid as a parameter.  Now that all methods that use
this decorator have been converted, the code can be simplified.

The _get_lock() method was updated to only look up an instance from the
db if necessary.  One of the unit tests had to be updated to account for
this.  These changes illustrate a good point about all of these
no-db-messaging patches.  Aside from silly typos, the types of bugs most
likely to occur due to these changes are due to code looking at stale
instance state since the state is being pulled out of the database less
often.  Hopefully I've caught everything, but it was worth pointing out
as something others should be aware of.

Part of blueprint no-db-messaging.

Change-Id: I02e6a5f2354f2e9ae8ea403b43359eb1bc8bae27

0 files changed, 0 insertions, 0 deletions