Add support for training classifiers with one class present

[tylerjthomas9]

Ref: #22

Should we have a warning when there is only one class?

Changes in this PR

• New fit, predict functions for CatBoostClassifier to handle the case when only one class is present.
• Bumped Aqua.jl version and added missing Compat entries
• Bumped actions and codecov in CI

[codecov]

Codecov Report

Attention: 6 lines in your changes are missing coverage. Please review.

Comparison is base (e7f56b7) 77.10% compared to head (b323d46) 75.55%.

Files	Patch %	Lines
src/mlj_catboostclassifier.jl	60.00%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #29      +/-   ##
==========================================
- Coverage   77.10%   75.55%   -1.56%     
==========================================
  Files           6        6              
  Lines         166      180      +14     
==========================================
+ Hits          128      136       +8     
- Misses         38       44       +6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ablaom]

This is kind of orthogonal, but FYI, to test row-based tables as well as column-based ones, we can now do this (with MLJTestInterface.jl 0.2.4):

Suggested change

	for data in [MLJTestInterface.make_binary(), MLJTestInterface.make_multiclass()]
	for data in [
	MLJTestInterface.make_binary(),
	MLJTestInterface.make_multiclass(),
	MLJTestInterface.make_binary(row_table=true),
	MLJTestInterface.make_multiclass(row_table=false),
	]

[ablaom]

Looks good to me.