PXC-4756: SST Post-processing Timeout in Large Instances (8.4)

[jaideepkarande]

https://perconadev.atlassian.net/browse/PXC-4756

Problem: In scenarios with large InnoDB Buffer Pools where
*innodb_numa_interleave* is enabled, when performing the *post-processing*
phase, the instance may take longer to start, potentially exceeding the
*timeout_threshold* of 300 seconds and causing the SST procedure to abort.

Cause:
The script *wsrep_sst_common* handles this logic. The value of
timeout_threshold is hard coded to 300 which maybe less for NUMA.
innodb_numa_interleave is a system variable that controls memory allocation
for InnoDB on Non-Uniform Memory Access (NUMA) systems.
When innodb_numa_interleave=ON, the mysqld startup process attempts to
interleave memory across all NUMA nodes, which can be useful for balancing
memory pressure but may lead to a slower start-up.
https://docs.percona.com/percona-server/5.7/performance/innodb_numa_support.html#version-specific-information

Solution:
Value of timeout_threshold now can be set via my.cnf file, parameter
post-processing-timeout in sst section.

[jaideepkarande]

Jenkins:
https://pxc.cd.percona.com/view/8.0%20parallel%20MTR/job/pxc-8.x-pipeline-parallel-mtr/612/