Low-risk correctness fixes: multi-peak param_bounds, crop(None), window index bound + test hardening

[gchure]

Summary

Three small, self-contained correctness/robustness fixes that change no behavior for valid input, plus hardening of the test suite's exception checks.

Fixes

I: multi-peak param_bounds validated the wrong peak's guess

In deconvolve_peaks, the global param_bounds feasibility check indexed the per-window initial-guess list p0 with a positive index (p0[key_inds[p]]), which always pointed at the first peak's parameters. For a window containing more than one peak, this validated the wrong peak's guess against the supplied bounds — risking a spurious ValueError or a missed one. The adjacent peak-specific block already uses the correct negative-index map paridx.

Fix: use p0[paridx[p]] so each peak is checked against its own guess; remove the now-dead key_inds. Single-peak windows are unaffected (the indices coincide), so existing test_generic_param_bounding / test_bounding are unchanged.

II: crop() with no/None window gave a cryptic error

crop() / crop(None) fell through to len(time_window) and raised TypeError: object of type 'NoneType' has no len().

Fix: add a leading guard that raises a clear ValueError instructing the caller to pass [start, end].

III: off-by-one in window-range index filter

_assign_windows filtered candidate indices with <= len(norm_int), admitting one out-of-range index. Harmless today (the value is only used via .isin) but incorrect.

Fix: use < len(norm_int). No observable output change.

Test hardening

Several exception tests used the try/except: assert True pattern with no assert False after the call, so they passed even when no exception (or the wrong one) was raised. Converted these to pytest.raises(...) intest_peak_fitting, test_deconvolve_peaks, and test_crop, and added regression tests for the multi-peak bound check and the crop(None) guard.

Risk

Low. No behavior change for valid input; the only newly-raised error (the crop(None) guard) replaces an existing crash with a clearer message.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.12%. Comparing base (391f9ee) to head (74ff25d).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #27   +/-   ##
=======================================
  Coverage   97.12%   97.12%           
=======================================
  Files           3        3           
  Lines         556      557    +1     
=======================================
+ Hits          540      541    +1     
  Misses         16       16           

☔ 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.