PQM: test failure

[Martin Pool]

> I ran into this one earlier today as well. As a simple "fix", I guess we
> could change to an RE match something like 0\.00[012]s for the time
> interval. It might still be slower under certain environments though.

I'm with Andrew; we should fix it to be time-independent so that it
doesn't depend on the variable timing.  It's easy for a process to be
delayed by over 100ms if the machine's busy.

-- 
Martin