Split current test_trainer into test_trainer and test_neuralnets

[DaniBodor]

test_trainer should tests core Trainer class functionality, e.g. missing network and dataset equivalence.

test_neuralnets should test each network from neuralnets subpackage. These are largely the tests that are currently using the model_base_test function. Also note that many neuralnets are currently not tested at all (see #284 )

[github-actions[bot]]

This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.