BUG: row_* functions in numba stats upcast float32 input to float64

[Nandinisingh07]

Reference Issues/PRs

Closes #3724

What does this implement/fix? Explain your changes.

The row_* functions in aeon/utils/numba/stats.py (row_mean,
row_count_mean_crossing, row_count_above_mean, row_median,
row_quantile25, row_quantile75, row_std, row_numba_min,
row_numba_max, row_slope, row_iqr, row_ppv) allocated their
output array with np.zeros(X.shape[0]), which always defaults to
float64 regardless of the input array's dtype. For float32 input
this silently upcast the result to float64, doubling memory usage
for the returned array.

This PR replaces the untyped allocation with one derived from the
input dtype, so float32 input now returns float32 output and
float64 input continues to return float64.

Also includes a cherry-picked fix for a related numba.TypingError
in Catch22._transform (_outlier_include), which surfaced while
verifying this change against downstream estimators that consume
these row functions (it failed to unify float64/float32 arrays
for outlier_series).

Does your contribution introduce a new dependency? If yes, which one?

No.

Any other comments?

Added a regression test, test_row_functions_preserve_dtype, in
aeon/utils/numba/tests/test_stats.py covering all affected
functions for both float32 and float64 input.

Verified no downstream regressions by running the fix against
CanonicalIntervalForestClassifier/Regressor, DrCIFClassifier/
Regressor, SupervisedTimeSeriesForest, RISTClassifier,
SevenNumberSummary, RandomIntervals, and SupervisedIntervals —
all pass. Also ran the full test_stats.py,
interval_based, and feature_based/catch22 test suites locally
(42 passed, 6 skipped as expected).

[aeon-actions-bot]

Thank you for contributing to aeon

I have added the following labels to this PR based on the title: [ bug ].
I have added the following labels to this PR based on the changes made: [ transformations ]. Feel free to change these if they do not properly represent the PR.

The Checks tab will show the status of our automated tests. You can click on individual test runs in the tab or "Details" in the panel below to see more information if there is a failure.

If our pre-commit code quality check fails, please run pre-commit locally and push the fixes to your PR branch.

Don't hesitate to ask questions on the aeon Discord channel if you have any.

PR CI actions

These checkboxes will add labels to enable or disable CI functionality for this PR. This may not take effect immediately, and a new commit may be required to run the new configuration.

• Run pre-commit checks for all files
• Run mypy typecheck tests
• Run all pytest tests and configurations
• Run all notebook example tests
• Run numba-disabled codecov tests
• Disable numba cache loading
• Regenerate expected results for testing
• Push an empty commit to re-run CI checks