fix: mask CMIP fill values (1e20) before monthly max/min aggregation

[rbeucher]

Problem

When data is loaded with decode_cf=False (as done throughout the pipeline to preserve lazy/Dask computation), fill values such as 1e20 are not automatically converted to NaN. This caused calculate_monthly_maximum and calculate_monthly_minimum to return 1e20 as the aggregated result when any fill value was present in the input data — most notably for tasmax and tasmin.

Fix

In both calculate_monthly_minimum and calculate_monthly_maximum in calc_utils.py:

1. Detect _FillValue / missing_value from both attrs and encoding
2. Mask those values via .where(da != fill_val) before calling .resample().max()/.min()
3. Preserves lazy computation — xarray.where is a lazy operation

Tests

Added two regression tests to test_derivations_calc_utils.py:

• test_masks_fill_values_in_attrs — verifies 1e20 in attrs is masked and does not appear in results
• test_masks_fill_values_in_encoding — verifies fill values in encoding are also handled

All 58 existing tests continue to pass.

[codecov]

Codecov Report

❌ Patch coverage is 77.77778% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 76.7%. Comparing base (9316f47) to head (f0d1543).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/access_moppy/derivations/calc_utils.py	77.8%	4 Missing ⚠️

❌ Your patch check has failed because the patch coverage (77.8%) is below the target coverage (90.0%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@          Coverage Diff          @@
##            main    #457   +/-   ##
=====================================
  Coverage   76.7%   76.7%           
=====================================
  Files         31      31           
  Lines       5927    5946   +19     
  Branches    1094    1097    +3     
=====================================
+ Hits        4546    4562   +16     
- Misses      1131    1135    +4     
+ Partials     250     249    -1     

Flag	Coverage Δ
unit	76.7% <77.8%> (+<0.1%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 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.