in the pantsd pre-fork context, we replace the

CpuPool

with

None

just before forking - but that leaves subsequent pantsd runs without a

CpuPool

which might cause the daemon to panic on the subsequent run?