Call unlock lockId before throwing an exception in acquireLock method

[maluchari]

In the cleans up path lockId is not returned in code for cleanup and hence becomes a no-op during unlock.

[rzhang10]

I have several points here:

• these 2 branches are when the lock state is waiting or is not required, this means other processes/threads must have hold the lock. So it's actually inaccurate if we try to release the locks on their behalf?
• it seems our hive doesn't enable concurrency currently, so seems this entire locking/unlocking on HMS might just have no effect to begin with. I doubt the original issue is really caused by this.

[maluchari]

Closing this as this seems to have been fixed upstream in this PR