[ENH] Refactor multiplication of classifier tests

Looking through the tests, I found that every classifier has a test file, and in it there is test_[classifiername]_on_unit_test_data and test_[classifiername]_on_basic_motions, with hundreds of lines of copy/paste. This is highly redundant and should be moved to a single file, using pytest.mark.parametrize and all_estimators.

#2257 makes a start how this could look like in the context of a classifier test suite refactor. I would suggest to branch off #2257 and follow the pattern. (assumes #2257 is merged. Until then, branch off #2257)

The recipe is as follows, for a given classifier:

1. find the file test_[classifier] or similar, containing tests test_[classifier]_on_unit_test_data and possibly test_[classifier]_on_basic_motions or similar.
2. move the parameter settings for the classifier into get_test_params, of the classifier class. Ensure it is the first element of the list returned. (also ensure there is no duplication with tests/_config, if there are still parameters there, also move them to get_test_params)
3. add the "expected outputs" from the test_[classifier] file to classification/tests/_expected_outputs (follow the pattern)
4. delete the file test_[classifier]
5. ensure the unit tests pass. They should, if this has been done 1:1

Classifiers to refactor:

• cboss
• muse
• tde
• weasel
• elastic_ensemble
• shape_dtw
• time_series_neighbors
• probability_threshold
• teaser
• catch22_classifier
• fresh_prince
• matrix_profile_classifier
• random_interval_classifier
• signature_classifier
• summary_classifier
• tsfresh_classifier
• hivecote_v1
• hivecote_v2
• cif
• dtc
• drcif
• rise
• stsf
• tsf
• arsenal
• rocket_classifier
• stc