If we want to respect the guarantee that the CB is called "no earlier than X", isn't ceil() more appropriate than round()?

@jlaine replied in #233 (comment)

Ok sorry for the duplicate message

So is the new test stable?

Yeah I think so

Alright, merged. Thank you! :)

@fantix got this CI failure this morning:

Could you please take another look at the test? Is it possible to make it any more reliable? If not, I suggest we relax it a bit or drop it.

(or maybe we should go with @jlaine suggestion to use ceil())

@1st1 ceil() may not fix this one - there's arbitrary time spent before the clock starts ticking. I think an absolute safe number would be 69, but it's probably not much meaningful anymore. For what is worth, maybe 69 for now?

Let's try it... Would the old code (without round()) fail on that?

@1st1 the old code may probably succeed with this test at 69 I think

Alright, I pushed e3b00b8

Oh, thanks for that! 🙇