change all for-loops in unit tests to parametrised tests

[lukashergt]

Instead of using for-loops in unit tests it would be better to use parametrised tests using @pytest.mark.parametrize. These decorators can also be stacked on top of each other in which case all combinations are performed (the equivalent of nested for-loops).

Using parametrised tests has the huge advantage that a failed test will tell which parametrisation exactly caused the issue, which would be lost in a for-loop.

Some of the unit tests have already been converted to parametrised tests. It would be good to extend that to all other tests that involve for-loops.

[williamjameshandley]

Partially implemented by @lukashergt in #109 and #99.

[williamjameshandley]

We should also move to more skips of e.g. fgivenx and astropy rather than checks #314