Make it easy to customize automl search data splitter

[dsherry]

Fix #1567

Usage: configure automl search to use a different number of folds (splits) in CV

random_state = 42
data_split = make_data_splitter(X, y, problem_type, n_splits=5, random_state=random_state)
automl = AutoMLSearch(problem_type=problem_type, data_split=data_split, random_state=random_state, ...)
...

Usage: disable shuffling

random_state = 42
data_split = make_data_splitter(X, y, problem_type, shuffle=False, random_state=random_state)
automl = AutoMLSearch(problem_type=problem_type, data_split=data_split, random_state=random_state, ...)
...

I suppose we could simply add n_splits and shuffle to AutoMLSearch.__init__, but I wanted to keep us flexible on this a while longer. We'll be adding more heuristics to the data splitting logic here at some point and its nice exposing that.

Hmm, perhaps we need to rename "data_split" to "data_splitter" as well.

[codecov]

Codecov Report

Merging #1568 (fcf7083) into main (22cd574) will increase coverage by 0.1%.
The diff coverage is 100.0%.

@@            Coverage Diff            @@
##             main    #1568     +/-   ##
=========================================
+ Coverage   100.0%   100.0%   +0.1%     
=========================================
  Files         239      240      +1     
  Lines       17593    17677     +84     
=========================================
+ Hits        17585    17669     +84     
  Misses          8        8             

Impacted Files	Coverage Δ
evalml/automl/__init__.py	100.0% <100.0%> (ø)
evalml/automl/automl_search.py	99.7% <100.0%> (-<0.1%)	⬇️
evalml/automl/utils.py	100.0% <100.0%> (ø)
evalml/tests/automl_tests/test_automl.py	100.0% <100.0%> (ø)
evalml/tests/automl_tests/test_automl_utils.py	100.0% <100.0%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 22cd574...fcf7083. Read the comment docs.

[bchen1116]

LGTM!

This change means we should file a separate issue to update our implementation for class_imbalance_data_check, specifically the cv_folds arg here. It defaults to 3 since that's the default data split n_folds, but with this change, I believe we should update AutoMLSearch to pass in the required param. Not blocking this PR though!

[dsherry]

@bchen1116 excellent point RE the class imbalance data check! I just filed as #1570

[freddyaboulton]

@dsherry Looks great!

[angela97lin]

LGTM! 😁