After N (1? 3? 5?) consecutive failed inits / loads we should try to destroy the database.

We could also try another init after the first one fails.

talo: Do you think it's reasonable to do this? Up until now it's not very common at all to fail initialization.

See histograms here:
https://uma.googleplex.com/p/chrome/histograms/?endDate=latest&dayCount=28&histograms=InProductHelp.Db.Init.AvailabilityStore%2CInProductHelp.Db.Init.EventStore%2CInProductHelp.Db.Load.AvailabilityStore%2CInProductHelp.Db.Load.EventStore&fixupData=true&showMax=true&filters=platform%2Ceq%2CA%2Cchannel%2Ceq%2C4%2Cisofficial%2Ceq%2CTrue&implicitFilters=isofficial

As of today, the numbers are <= 0.01 for both init and load of both the event store and the availability store.