Since SERVER-124364 in SERVER-124363 we interrupt and retry DB and collection filtering metadata refreshes after we enable/disable the Authoritative Shards CRUD authoritative shards. Behind the scenes this signals the refresh's cancellation token, but it does not wait for the refresh to acknowledge it (the refresh should quit at the next interruption point, but it can still perform work before that).

This issue is to evaluate if we should also drain the refreshes - i.e. wait for them to finish - after the FCV transition.