Description and related issues

Cleanup work for the trialization/tokenization pipeline.

Better separation of trialization logic. Rename MarkTokenizer to TrialsManager, and let this handle the nwb_content.add_trial(…) operations. The TrialsManager uses the Tokenizer classes to construct a list of trials, but the Tokenizer classes now do not work directly on the nwb_content.
Among the tokenizer classes, move more common actions to BaseTokenizer (renamed from StimulusTokenizer). Refactor features into separate methods.
~(potentially incorrect) hard-code a mark threshold of 0.1 for WNTokenizer, instead of using 0.3 as specified in the wn2.yaml stimulus metadata. No justification for this value 0.1. Empirically this leads to a successful identification of all 60 stim onsets in the test block RVG16_B01.~ —————— EDIT: this is now commented out, so the behavior is unchanged.
Other than ^this^ (change in WN mark threshold; which is now suppressed too), I believe that tokenizer behavior — the intended behavior, at least — is unchanged.

Didn't resolve the number of stim onset mismatch issues (for the WN and TIMIT tokenizers), but set the stage up…

Note added: the TIMIT tokenizer error is probably due to the test block being incomplete. This should be tested/fixed separately from this PR, once a test dataset is ready.

Closes #(issue number)

Checklist:

[ ] All tests pass on catscan: run pytest --basetemp=tmp -sv -n 8 tests on catscan from the root directory
[x] If needed, docs have been update: docs/source has been updated for any added, moved, or removed files
[x] Docs build with no errors: run make clean & make html from the docs folder
[ ] No python formatting errors: run flake8 nsds_lab_to_nwb tests from the root directory