Replay-blockchain option not purging state code cache

[eosusa]

When attempting to kick off a --replay-blockchain on a 3.2.4 node that crashed and left dirty db, I get the following error and it will not proceed:

info  2023-07-13T16:11:29.593 nodeos    chain_plugin.cpp:966          plugin_initialize    ] Starting fresh blockchain state using genesis state extracted from blocks.log.
info  2023-07-13T16:11:29.603 nodeos    block_log.cpp:284             open                 ] Log is nonempty
info  2023-07-13T16:11:29.603 nodeos    block_log.cpp:317             open                 ] Index is nonempty
warn  2023-07-13T16:11:29.604 nodeos    chain_plugin.cpp:1251         plugin_initialize    ] 3060000 database_exception: Database exception
code cache is dirty
    {}
    nodeos  code_cache.cpp:234 code_cache_base

error 2023-07-13T16:11:29.604 nodeos    main.cpp:161                  main                 ] 3060000 database_exception: Database exception
code cache is dirty
    {}
    nodeos  code_cache.cpp:234 code_cache_base
rethrow
    {}
    nodeos  chain_plugin.cpp:1251 plugin_initialize

Manually purging the state directory before the replay did resolve the issue (but ironically led me to the new "feature" where a missing state folder wont delete ship logs and errors out with generic cannot write to statehist logs error)

[spoonincode]

I think it might make sense to silently remove the code cache if it's dirty.