`Categorical categories must be unique` ValueError during score_lambda_path

[codyheiser]

Random error during lambda path scoring with Parallel backend. Traceback below:

Training dropkick with alphas:
        [0.1]
[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.
[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:   10.1s remaining:   15.2s
Traceback (most recent call last):
  File "/home/cody/.virtualenvs/sc/bin/dropkick", line 33, in <module>
    sys.exit(load_entry_point('dropkick', 'console_scripts', 'dropkick')())
  File "/home/cody/git/dropkick/dropkick/__main__.py", line 281, in main
    args.func(args)
  File "/home/cody/git/dropkick/dropkick/__main__.py", line 53, in run
    _ = dropkick(
  File "/home/cody/git/dropkick/dropkick/api.py", line 551, in dropkick
    rc_.fit(adata=a, y=y, n_hvgs=n_hvgs)
  File "/home/cody/git/dropkick/dropkick/logistic.py", line 268, in fit
    cv_scores, _hvgs = _score_lambda_path(
  File "/home/cody/git/dropkick/dropkick/util.py", line 68, in _score_lambda_path
    *Parallel(n_jobs=n_jobs, verbose=verbose, backend="threading")(
  File "/home/cody/.virtualenvs/sc/lib/python3.8/site-packages/joblib/parallel.py", line 1042, in __call__
    self.retrieve()
  File "/home/cody/.virtualenvs/sc/lib/python3.8/site-packages/joblib/parallel.py", line 921, in retrieve

...

ValueError: Categorical categories must be unique

Ran again immediately after this error on same file with no problem; wondering if it was a stochastic label conflict during Parallel training. Keep notes for potential future issues.