tests: fix tests and use layers from nc_spm_basic dataset

[petrasovaa]

I am trying to revise the dataset used in tests with the goal of creating a new dedicated test dataset based on the nc spm basic dataset. Right now the tests use the combination of both basic and full dataset. As part of this I fixed many tests that didn't make any sense and were not testing anything.

There will be more, but this is already enough to review.

[echoix]

I know it's not exactly related, but you might be interested to know that. I'm really near to have finished adding a new decorator for setting expected failures to all failing test functions (not files) on Windows. Then I will be able to set the Windows tests to require all tests passing. When tests are unexpected successes, we'll know, and if it is caused by a real fix, we remove the decorator and that test cannot fail again.

I'm only a couple days away from it, as I don't know exactly what to do with straight up errors in the setup class, and still need to review if my last round of adding decorators handles all of the failing tests.

We shouldn't be stepping on our foots, but you might still want to be aware of it.