As part of a HELP-21209 ticket discussion, we observed that history store file size doesn't get reduced in PSA architecture once the standby catches up with the primary after a crash.
As part of the checkpoint, the checkpoint garbage collection removes the obsolete pages that are part of any table and later these removed pages will be cleared and space gets reclaimed. WiredTiger checkpoint process skips the tables as part of the checkpoint that doesn't have any modifications since the last checkpoint (exceptions are there).
The test that is described in the help ticket doesn't let any modifications happen to the WiredTiger history store table once the standby catches up with the primary. As part of WT checkpoint, the history store file is always skipped to be part of the checkpoint as there are no modifications and it led to never removal of obsolete pages that are part of the history store table.