Limit number of diffs per worker process in server

[Mr0grog]

This adds a new MAX_DIFFS_PER_WORKER environment variable that limits the number of diffs performed by a single worker. After a worker process performs this many diffs, it is shut down and replaced with a fresh process.

This is an ugly first cut and needs a lot of cleaning up. It’s probably time to address the longstanding to-do here:

web-monitoring-diff/web_monitoring_diff/server/server.py

Lines 503 to 507 in b3cf524

	# TODO: we should split out all the management of the executor and diffing
	# (so this, get_diff_executor, caller, etc.) into a separate object owned
	# by the server so we don't need weird bits checking the server's
	# `terminating` state and so that all the parts are grouped together.
	async def diff(self, func, a, b, params, tries=2):

This also doesn't really account for things on a per worker basis — it just restarts the pool after DIFFER_PARALLELISM * MAX_DIFFS_PER_WORKER diffs. But newer versions of Python have an API that does this right and which we should eventually switch to when compatible (although it seems like there maybe be some bugs in it: #202 (comment)).

Fixes #202.

[Mr0grog]

Rebased to merge release notes cleanly. This is messy, but I’m thinking refactoring diff pool management may be big enough that it should be done in a second followup PR.