Unify numerical comparison across testing framework

[davidrudlstorfer]

Closes #184

[davidrudlstorfer]

1. The cosserat tests are currently failing because the global tolerance is too tight. Should we redo the results?

2. With the current changes I introduced the numerical comparison into the test framework. Should we instead introduce it into the general MeshPy utils and also use it in MeshPy?

3. Should we globally "reconfigure" the tolerances? my.eps?

[davidrudlstorfer]

Looks like the cosserat tests were created on macOS because there they succeed :D

[isteinbrecher]

I really like how this improves the testing comparisons. (Actually quire strange what constructs could be found here 😂)

[isteinbrecher]

Can we use the same global tolerances as introduced in #382?

[isteinbrecher]

• The cosserat tests are currently failing because the global tolerance is too tight. Should we redo the results?

What exactly do you mean with "redo the results"?

• With the current changes I introduced the numerical comparison into the test framework. Should we instead introduce it into the general MeshPy utils and also use it in MeshPy?

Without thinking to long about that, I would say yes, that would be a nice follow up.

• Should we globally "reconfigure" the tolerances? my.eps?

You mean to unify them or lower the tolerances?