Use the median of the delta instead of min for time freq inference

[chengzhuzhang]

Description

Using the median of delta is more tolerant when handling small irregularities (e.g., small drift in timestamps).
min() can be too sensitive and lead to incorrect frequency inference (e.g., #760 which has been tested with this PR)

• Closes [Enhancement]: Improve inference logic for time frequncy #762

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules

If applicable:

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass with my changes (locally and CI/CD build)
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have noted that this is a breaking change for a major release (fix or feature that would cause existing functionality to not work as expected)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (05ad2f2) to head (b3c08bb).

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #768   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           16        16           
  Lines         1705      1706    +1     
=========================================
+ Hits          1705      1706    +1     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[tomvothecoder]

Thank you @chengzhuzhang for getting to this before me! This PR looks ready for review? If so, I will review and merge if it is good to go.

[Copilot]

Pull Request Overview

This PR refactors the time frequency inference logic by replacing the minimum time delta with the median delta, making it more robust against small timestamp irregularities.

• Replaces min() with median() for calculating the time delta.
• Adds a comment to ensure that the time coordinates are sorted before processing.
• Updates conditional checks for "hour", "day", "month", and "year" frequencies.

Comments suppressed due to low confidence (1)

xcdat/temporal.py:2108

• The comment indicates time_coords should be sorted; to avoid unexpected behavior due to unsorted input, consider enforcing or explicitly sorting time_coords before computing diffs.

time_deltas = np.diff(time_coords.values).astype("timedelta64[ns]")

[tomvothecoder]

Note, in the docstring I cited the native xarray.infer_freq() method not working for all cases, which is why I implemented _infer_freq():

The native ``xr.infer_freq()`` method does not work for all cases
because the frequency can be irregular (e.g., different hour
measurements), which ends up returning None.

[tomvothecoder]

Hi @chengzhuzhang, this PR looks good to me.

I added direct tests for _infer_freq() function to ensure it is working as intended.

[chengzhuzhang]

@tomvothecoder Thank for a review! I think it is ready to merge.
@pochedls thanks for the suggestion on this change!