Rebalance handling: reduce time holding write lock on signal

[gterzian]

This does not fix any open issue, it is rather something that was itching while reading the code.

I'm wondering if you want to hold the write lock on the rebalance signal while handling the rebalance, so I've tried to minimize the time the lock is held.

Current logic appears to be:

• set the signal back to None, but only if there wasn't an error in processor.table.update(), so I've tried to keep that logic intact.
• Also, since the lock is currently held while processor.table.update() is called, a new signal cannot replace it until the lock is dropped. Tried to keep that part too.

With these changes, the base case--when there is no error handling the signal--will only see the write lock acquired once, and held for so long as it takes to call take on an Option<_>.

[rtyler]

:confused: I am really not sure why CI is not running for this pull request :confused:

[gterzian]

I am really not sure why CI is not running for this pull request

I think it's because it's from a fork. See https://docs.github.com/en/actions/managing-workflow-runs/approving-workflow-runs-from-public-forks

cc @rtyler

thanks for the merge!

[rtyler]

@gterzian I believe that there was likely a GitHub Actions outage at the time. The prompt to Approve and Run did show up again over the weekend :shrug: