Phase column and phase_statistics on phased X/mR (#144, PR 3)

[cnicholas]

Part of #144 (PR 3 of 4). Stacked on #150: the base is feat/phased-xbar-s, so the diff shows only this change.

Summary

• What: phased X and mR charts gain what phased Xbar/S got in Phased limits on Xbar and S charts (#144, PR 2) #150: a phase column (the lane label) on the chart data and in chart_table, and result.phase_statistics(chart) with one row per phase (phase, points, center, mR_bar, lpl, upl). On a stratified phased chart (by=['factor 1'], phased=True) the table is per stratum: phase_statistics(chart, stratum=...) or focus(stratum).phase_statistics(chart). The companion mR now reports single_point_phases, which only the solo mR did.
• Why: today a phased X/mR chart computes each phase's centre, mR-bar and limits and then keeps them only on the rows. The per-phase numbers are the answer the chart gives, and the same accessor should work on every phased chart.
• Scope: output additions only. The phased arithmetic, row order, limits and signals are unchanged; phased=False charts are untouched.

Contract / invariants

• Defaults unchanged — unphased X/mR output untouched (the new column and payload key exist only when phased=True); validation/xbar_s_sweep.py identical (0 differing lines)
• Residuals unaffected
• Row/index alignment preserved — no rows added, dropped or reordered; lane boundaries unchanged
• Output schema compatible — additive: phase column after rsg, phases payload key, single_point_phases on the companion mR. Golden masters untouched (none of the scenarios is phased)
• Pinned error strings unchanged

Behaviour changes

• Phased X/mR chart data: ['time', 'rsg', 'phase', value, 'center', 'lpl', 'upl', 'beyond_limits', 'obs_id'] (stratified: the by column before phase); chart_table shows phase after subgroup.
• phase_statistics(chart, stratum=None) works on phased X/mR; on a stratified one it asks for a stratum and names the available ones.
• Companion mR metadata carries single_point_phases.

Methodology

• No methodology change — the per-phase values were already computed (10-12 manual Eqs 12-1, 12-5, 12-10, 12-11 per phase); they are now reported

Tests

• tests/test_phased_limits.py::TestPhaseColumnAndStatistics (5): the phase column equals the lane label and each phase has one centre; the X and mR tables have one row per phase in display order, points sum to the rows, each row's centre and limits match the chart rows, the mR centre is its mR-bar with LPL 0 and the same mR-bar as X; the companion mR carries single_point_phases equal to X's and to the count of phases under two points; stratified tables by stratum equal focus(stratum).phase_statistics, missing or unknown strata are refused; unphased charts have no phase column and refuse phase_statistics, and a stratum on an unstratified chart is refused
• tests/test_phased_xbar_s.py: the X/mR refusal test became the stratum refusal
• pytest tests/ — 2633 passed, 7 skipped
• ruff check .
• mypy processbehavior — 14 errors, all pre-existing, none in the touched code
• python validation/e2e_bishop_report.py — 1547 passed, 0 failed, 4 pending; docs/reference/validation.md unchanged

Notes

• Docs: api.md "Phases" covers X/mR and the stratum argument; CHANGELOG entries from Phased limits on Xbar and S charts (#144, PR 2) #150 extended, plus a Fixed line for the companion mR.
• Follow-up: PR 4, ordered Categoricals on X/mR lane order.