Don't fail for > 1 walsenders in 019_replslot_limit, add debug messages.

Don't fail for > 1 walsenders in 019_replslot_limit, add debug messages.

So far the first of the retries introduced in f28bf667f60 resolves the
issue. But I (Andres) am still suspicious that the start of the failures might
indicate a problem.

To reduce noise, stop reporting a failure if a retry resolves the problem. To
allow figuring out what causes the slow slot drop, add a few more debug
messages to ReplicationSlotDropPtr.

See also commit afdeff10526, fe0972ee5e6 and f28bf667f60.

Discussion: https://postgr.es/m/20220327213219.smdvfkq2fl74flow@alap3.anarazel.de