Sometimes we get stuck on db compaction, and the CL re-issues the "same" command to us multiple times. Each request get stuck on the same place, in the middle of the handler.

This changes makes it so we do not reprocess the same payload, but instead detects it early.