It is not a wholesale search and replace, I have checked each change. There are a lot of places in the tree where monotonic is not appropriate, and those should not be included in the patch. But there is always the possibility that I've missed something, and it's good that there are reviewers.

Granted, some of the changes in this patch might not make a big difference, but they should not hurt either. To me, using monotonic is the right thing to use when dealing with time differences/delays, unless one needs an actual predictable time value anchored at the epoch at the same time for some other purposes. I don't of course insist on that point of view, and would be interested to hear other opinions and if someone can point out flaws/dangers in mine.