Pseudo-semisync: Async replication with Slave acknowledgement

Observers can interpret these ACKs as a progress-reporting (MDEV-21322) form of slave heartbeating.

Questions to answer:

• What is the incentive? For up-to-date information, why wouldn’t the Observers directly ping SHOW SLAVE STATUS on the slaves themselves?
• From MDEV-21322:

  [...] treat rpl_semi_sync_master_timeout=0 as a special case where transactions do not await an ACK, but still report progress from the ack_thread on replica reply.

  Is it really ergonomic to effectively disable semisync for the whole master? Maybe this is better as a slave-side per-connection config (part of MDEV-40941).

• From MDEV-21322:

  Per design, when rpl_semi_sync_master_timeout=0, than rpl_semi_sync_master_status, if enabled earlier, should stay ON , instead of switching off to async and incrementing the counter updates.

  Would it be clearer to present itself as OFF, matching the current behavior, especially to remind that rpl_semi_sync_master_timeout=0 did and will not mean "wait forever"?