Fix background worker not restarting after crash-and-restart cycle.
Previously, if a background worker crashed (e.g., due to a SIGKILL) and
the server restarted due to restart_after_crash being enabled,
the worker was not restarted as expected. Background workers without
the never-restart flag should automatically restart in this case.
This issue was introduced in commit 
28a520c0b77, which failed to reset
the rw_pid field in the RegisteredBgWorker struct for the crashed worker.
This commit fixes the problem by resetting rw_pid for all eligible
background workers during the crash-and-restart cycle.
Back-patched to v18, where the bug was introduced.
Bug fix patches were proposed by Andrey Rudometov and ChangAo Chen,
but this commit uses a different approach.
Reported-by: Andrey Rudometov <[email protected]>
Reported-by: ChangAo Chen <[email protected]>
Author: Andrey Rudometov <
[email protected]>
Author: ChangAo Chen <
cca5507@qq.com>
Co-authored-by: Fujii Masao <[email protected]>
Reviewed-by: ChangAo Chen <[email protected]>
Reviewed-by: Shveta Malik <[email protected]>
Discussion: https://postgr.es/m/CAF6JsWiO=i24qYitWe6ns1sXqcL86rYxdyU+pNYk-WueKPSySg@mail.gmail.com
Discussion: https://postgr.es/m/
[email protected]
Backpatch-through: 18