Fix RotD calculation

[lispandfound]

Found due to work from Alan. The old code essentially calculated the wrong side of the median in the RotD50 calculation, resulting in RotD50 being biased high. The fix is a small one to use bog-standard sorting to calculate median, which is not only simpler but more correct. I have also added a bunch more rotd calculation tests, including hypothesis tests that assert:

1. rotd100 / rotd50 remains within floating point epsilon of sqrt(2) (guarantee is 1e-12),
2. For a special class of records (linearly polarised records), the sqrt(2) bound is precisely attained,
3. For all pairs of rotd_0 and rotd_90 the parallel codepath is exactly identical to the single-threaded codepath. I plan to remove the parallel code at a later date but in the meantime this is an important assertion.

By "hypothesis test" I mean testing in the form of the hypothesis library: automatically testing a wide variety of cases and then crucially shrinking to the simplest possible failure case.

Line count is high because I bumped the uv and cargo lock files. Real pull request is +300ish lines, mostly tests and some minor ruff fixes.

[Copilot]

Pull request overview

Fixes the Rust RotD50 (median) computation to avoid biased-high results by switching to a fully sorted median selection, and strengthens correctness guarantees with additional property-based RotD tests. The PR also includes several small Python-side typing/lint cleanups plus dependency / lockfile updates.

Changes:

• Correct RotD50 calculation in src-rust/rotd.rs by sorting all 180 rotated peaks and taking the proper middle-pair average.
• Add extensive Rust proptest coverage around RotD invariants (sqrt(2) bound, tightness for linearly polarised records, and serial vs parallel equivalence).
• Minor Python refactors/cleanups (safer defaults for CLI + API parameters, metadata loop simplifications, typing suppression updates) and dependency bumps.

Reviewed changes

Copilot reviewed 11 out of 13 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src-rust/rotd.rs	Fixes RotD median computation and adds property-based tests for invariants and parallel/serial agreement.
Cargo.toml	Updates Rust deps (ndarray/numpy/pyo3) and adds proptest for new tests.
Cargo.lock	Lockfile updates to reflect Rust dependency changes.
IM/im_calculation.py	Avoids mutable default argument for ims_list and updates type-suppression comments.
IM/scripts/calculate_ims.py	Avoids mutable default list options by defaulting periods/frequencies to None and filling them at runtime.
IM/im_reader.py	Minor cleanup to use loop variables when setting coordinate/IM metadata descriptions.
IM/ims.py	Updates type-suppression comment style and removes an unnecessary suppression on os.environ.update.
IM/snr_calculation.py	Updates type-suppression comment style for DataFrame column assignment.
tests/test_ims.py	Small test cleanups, updated type-suppression comment style, and minor dtype tightening in one hypothesis case.
tests/test_im_reader.py	Minor iteration cleanup (for im_name in IM_METADATA vs .keys()).
pyproject.toml	Bumps minimum supported Python version to 3.12.
.github/workflows/pr-review.yml	Removes an “AI Code Review” reusable-workflow integration.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.