Re: WIP: Failover Slots

In particular, for the checkpoint issue I landed up moving CheckPointReplicationSlots to occur at the start of a checkpoint, before writing WAL is prohibited. As the comments note it's just a convenient place and time to do it anyway. That means it has to be called separately at a restartpoint, but I don't think that's a biggie.

The tests for this took me quite a while, much (much) longer than the code changes.

I split the patch up a bit more too so individual changes are more logically grouped and clearer. I expect it'd be mostly or entirely squashed for commit.