Tolerance in tests ?

[Abelarm]

Hi @phoebe-p,

I keep getting annoying messages from Github about test failing in github action, for my fork of solcore.
All of them are because the tolerance is way too low.
I checked and also the test for solcore they usually fail for the same reason

I wanted to ask if is it by design that solcore need this strict tolerance or we can relax it a bit to not make the test fail?

Examples:

[phoebe-p]

Sorry, missed this somehow. We can definitely increase the tolerance on the tests, this is a silly reason to have failed tests.

[Abelarm]

No worries, what's a good tolerance we like?

[phoebe-p]

I've increased the tolerance for some tests I have seen failing due to this problem in #278. There are probably others that I missed. There were a couple where no tolerance was set in approx at all (so it's using the default of 1e-12 absolute/1e-6 relative), which I set to a relative tolerance of 1e-5. There were some other tests where a tolerance of 1e-4 was set which sometimes failed, changed those to 1e-3. let me know if it keeps happening with other tests! I will also keep an eye on the daily scheduled tests to see if there are other failures related to this (now that they are working again not failing for other reasons in the build system...)