The Platform team is investigating the feasibility of mocking the modulestore interface with the goal of saving time on unit test runs:
A good portion of the time spent in tests actually comes from the events that trigger off of course publish signals, rather than the modulestore operations themselves. Removing these (or only selectively turning these on where needed) will reduce test run time significantly. This will also help make tests more predictable, so that adding a new publish listener won't break a bunch of query counting tests. It's similar in principle to what we already have in place for cache isolation.