Add value counts aggregation to select.group_by

[ebhills]

Summary

• add a counts aggregation to select.group_by
• return each distinct value and its frequency as a dictionary
• convert dictionary keys to JSON-safe strings
• represent missing values as null and booleans as lowercase true or false
• support existing per-column output renaming behavior

Why

Grouped recipes sometimes need the distribution of values within each group rather than only a scalar count or a list of the original values. Without this aggregation, recipes need additional transformations to construct a value-to-frequency mapping.

Impact

Recipes can now request value counts directly:

wrangles:
  - select.group_by:
      by: Group
      counts:
        - Selection: Selection Counts
      auto_rename_columns: false

The resulting Selection Counts value is a JSON-safe dictionary such as {"Primary": 2, "None": 1}.

Validation

• complete TestGroupBy class: 30 passed
• grouped smoke coverage for strings, nulls, and booleans passed
• git diff --check passed
• two existing Python invalid-escape SyntaxWarning warnings were emitted; no test failures

[Copilot]

Pull request overview

Adds a new counts aggregation option to select.group_by so grouped recipes can return per-value frequency distributions (as JSON-safe dictionaries) instead of only scalar aggregates.

Changes:

• Extend select.group_by to support a counts aggregation that returns {value_as_string: frequency} per group.
• Normalize keys for JSON-safety (string conversion, lowercase boolean strings, and a "null" key for missing values).
• Add a unit test validating the new aggregation and per-column output renaming with auto_rename_columns: false.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
wrangles/recipe_wrangles/select.py	Adds schema documentation and implements the counts aggregation using value_counts(dropna=False) with JSON-safe key normalization.
tests/recipes/wrangles/test_select.py	Adds a test covering counts output structure and column renaming behavior.