swim protocol seems to busy spin sometimes

[carns]

I don't have a solid reproducer for this, but I wanted to record what I have.

If you start a single bedrock daemon with ssg and swim enabled, then the bedrock daemon consumes 100% CPU forever after servicing unrelated provider RPCs (like from mochi-quintain).

If I add the following to the bedrock json to keep ssg but disable swim, then the bedrock daemon wilil return to ~0 CPU consumption when idle:

       "swim": {
                   "disabled": true
                    }

I'm not sure why enabling/disabling swim would make a difference in this case, since there is only one process in the group. There shouldn't be anything to monitor in that case?