Hi Michael,

Thanks for taking a look!  Ah I see, I thought that refining on the PID of
a process only filtered the buffer to show the process and it's direct
children but looking at 'proced-filter-children' this doesn't appear to be
the case.

I've attached a patch which changes the refinement tests to use the simpler
CPU refinement behaviour (keep everything with %CPU >= process at point),
which appears to work well.

On the downside, whilst running the new tests I noticed an occasional flake
with the 'proced-update-preserves-pid-at-point-test' test.  Unfortunately I
still can't find the root cause after debugging it for a while so I've
marked it as unstable in the patch for now as well.

Thanks, Laurence