[Parity] Close final standalone evidence

[suraj-ranganath]

Close the final standalone parity phase by reclassifying completed documentation matrix rows, adding closeout evidence notes, documenting time-frequency and movie workflow boundaries, and tightening tests for closeout status and HDF5 timing semantics.

Closes #165

[claude]

Claude finished @suraj-ranganath's task in 3m 5s —— View job

---

Code review — #173 [Parity] Close final standalone evidence

• Gather context (diff, repo instructions)
• Review correctness & regressions
• Review EEGLAB parity & data structures
• Review tests & docs

---

Code review

• Overall assessment: Safe to merge.
• Highest-risk area: Doc code examples (verified accurate) and the HDF5 fixture timing change.
• Merge recommendation: Safe to merge.

This is a docs/evidence closeout PR (parity matrix reclassification, two new doc sections, tightened tests). I verified the substantive risks below.

Blocking

None.

Important

None.

Nits

None.

Test gaps

None. The new tests in tests/test_eeglab_final_parity_matrix.py cleanly encode the closeout invariants:

• test_final_closeout_has_no_unimplemented_or_documentation_gap_rows — confirmed no port/docs_gap rows remain (all 31 rows are now implemented/consolidated/partial/optional_dependency/matlab_runtime_skip/stale_skip).
• test_final_partial_rows_are_explicit_backend_boundaries — both partial rows (dipfit-dipole-fitting line 616, optional-fieldtrip-study-neighbors line 952) contain "limitation"/"boundary" in their rationale/test_notes, so the assertion holds.
• The taxonomy assertion was correctly relaxed from {"optional_dependency", "docs_gap"} <= statuses to "optional_dependency" in statuses, consistent with docs_gap now being eliminated.

EEGLAB parity notes

• HDF5 fixture timing (tests/test_pop_loadset_h5.py): Changing xmax from 1.0 to 0.998 and times to np.linspace(-1, 0.998, 1000) makes the fixture internally self-consistent with EEGLAB timing semantics: xmax = xmin + (pnts-1)/srate = -1 + 999/500 = 0.998 at srate=500, pnts=1000. This is a correct parity improvement, not a tolerance relaxation. ✅
• Doc code examples (preprocessing_pipeline.rst): All four referenced wrappers are exported from eegprep and their signatures accept the documented arguments — pop_newtimef(EEG, typeproc=, num=, return_com=), pop_newcrossf(EEG, typeproc=, num1=, num2=, return_com=), pop_spectopo(EEG, dataflag, timerange=, return_com=), pop_erpimage(EEG, typeplot=, index=, return_com=). Examples will run as written. ✅
• Matrix eegprep_equivalent additions: The doc paths newly attached to docs-event-processing-tutorials, docs-source-reconstruction-tutorials, and docs-time-frequency-and-movie-tutorials all point to existing files already present as docs_targets. ✅
• RST validity: New section headings have underlines ≥ title length; the :ref:\preprocessing_pipeline`target invisual_parity.rstresolves to the existing.. _preprocessing_pipeline:` label. No build-break risk. ✅

The movie/visual-workflow and time-frequency boundary notes are accurately framed as explicit tooling/future-work boundaries rather than claiming unimplemented features, which matches the AGENTS.md guidance against documenting intentions as implemented behavior.

Checked for correctness bugs, EEGLAB parity, data-structure compatibility, changed-behavior tests, and CLAUDE.md/AGENTS.md compliance. Note: I validated the test invariants by inspecting the matrix data directly; sandbox restrictions prevented me from executing the suite, but all four assertions hold against the committed JSON.