Differential expression genes: test to use?

I have a question on the computation of differential expression genes with the scanpy function scanpy.tl.rank_genes_groups().

There is a way to understand which is the correct statistical test to use when computing DEGs?
By default the function uses the wilcoxon method, but is it uncorrect to change the test in the function with, for example, t-test_overestim_var? Is it up to the operator to choose the test or there some test that is possible to calculate first that can help in choosing the right statistical test?
The t-test_overestim_var method test, when is correct to use?

Thank you!