db522b542e
A few regtests continue to regularly fail in highly loaded VMs because they have very short timeouts. Actually the goal of running with short timeouts was to make sure we do not uselessly wait during tests designed to trigger them, but these timeouts here are never supposed to fire at all, so they don't need to be kept in the 15-20ms range. They do not pose any issue on any regular machine, but VMs are often suffering from huge time jumps and cannot always produce responses in that short of a time. Just like with commit ce6fc25b1 ("REGTEST: increase timeouts on the seamless-reload test"), let's raise these short timeouts to 1 second. A few other ones remain set to 150-200ms and do not seem to cause any issue. Some are actually expected to trigger so let's not touch them for now.