fix test flakes

[almogdepaz]

assert spend is in mempool before farming blocks

reduce farmed blocks in tests

reduce timeouts

make tests consistent

[lgtm-com]

This pull request introduces 1 alert when merging 6695527 into 9a0a2b1 - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

[altendky]

The reduced timeouts concern me in terms of CI runners liking to just randomly stop running our stuff for awhile, at least on macOS in my previous experiences, but otherwise I'm good with this. I can give it a big rerun blast in another PR to see if my concerns are well founded or not. If not I'm happy to merge this into the coverage branch.

[altendky]

Also... thanks!

[almogdepaz]

The reduced timeouts concern me in terms of CI runners liking to just randomly stop running our stuff for awhile, at least on macOS in my previous experiences, but otherwise I'm good with this. I can give it a big rerun blast in another PR to see if my concerns are well founded or not. If not I'm happy to merge this into the coverage branch.

i dont mind changing back but i feel like this is just being too defensive so we can avoid CI failures where the real goal is to find bugs/slowdowns
i think the timeouts here are just a result of hiding the real issue/copy paste from other tests
there is no reason for 20 sec, can we run this like 100 times and see? i hate giving long timeouts it just hides stuff

[altendky]

Testing robustness in #13217. That is running only the wallet.rpc job and creating 60x duplicates. That hits 240 jobs in the linux and windows workflows where 255 (or 256) is the limit per matrix.

If that works out, ok. I both agree with disliking long timeouts and also have spent too much time dealing with macOS runners out of the blue doing nothing at all for 30 seconds at a time. Anyways, we'll see what the run above says.