Make file purge logic less aggressive

Currently we have purge logic for log file directories and secret files that runs on host startup. We should consider doing these purge operations less often, perhaps on a periodic background thread not tied to host startup. This might also yield startup time improvements.

No files, tests, or entry points are identified in the issue. Start by locating the host-startup purge logic for log directories and secret files, then trace how often it runs and what scheduling options exist. Done should mean purge work is no longer tied to every startup while log and secret cleanup still occurs as intended and startup time improves or does not regress.

Written by the indexing model from the issue text.