fix(plotting): lane-chart line breaks at lane boundaries (#121); docs: series length + derived variables

[cnicholas]

Summary

• What: on a lane chart (an X or mR chart with factors and by=[]), the connecting line is now its own trace with a gap at each lane boundary, drawn under the markers. Plus two user-guide pages.
• Why: Break the connecting line at lane boundaries on stratified X/mR charts #121. The last point of one subgroup was joined to the first point of the next, so every boundary drew a steep rise or drop that was only the sort order; zoomed to a few lanes it read as a collapse between each organisation. The docs pages state the series-length position from Series length (T) as an axis in design-state admissibility — and a saturation question #114 (with the n = 3..30 table from validation/short_series_bands.py and credit to @rabujamra) and the derived-variables contract from fix(derivations): evaluate/validate never raise; non-finite results are violations; construction validates #126. Neither existed anywhere a user would look.
• Scope: plotting/renderers.py (main-trace block, two helpers); tests/test_lane_connectors.py; docs/user-guide/series-length.md, docs/user-guide/derived-variables.md, docs/myst.yml, docs/getting-started/key-concepts.md; CHANGELOG.

Contract / invariants

• Defaults unchanged — single-series charts and faceted charts keep the original single lines+markers trace with no extra trace (pinned). validation/e2e_bishop_report.py exits 0.
• Residuals unaffected — plotting only.
• Row/index alignment preserved — the markers trace is the same points in the same order; the line trace repeats the boundary x for the gap entry so x arrays contain no nulls (the numeric-x axis test still holds).
• Output schema compatible — the markers trace keeps its name, hovertemplate, customdata, legend and signal highlighting.
• Pinned error strings unchanged.

Behaviour changes

• Lane charts gain a lines trace named <chart> (lanes) (hoverinfo='skip', showlegend=False) and the markers trace's mode becomes markers. Visually: the same points, the same limits, and no segment crossing a lane boundary.

Methodology

• No methodology change — rendering and documentation.
• The series-length page's "position" section paraphrases Tom's 2026-09-10 email; he should read it once.

Tests

tests/test_lane_connectors.py:

• one gap per boundary, at the lane-start positions; point counts unchanged

• no drawn segment spans two lanes (lanes 100 apart, segments never jump)

• markers keep hover; the line trace has none and no legend entry

• single-series chart unchanged; faceted chart unchanged

• pytest tests/ — 2426 passed, 10 skipped

• ruff check . — clean

• mypy processbehavior/plotting/renderers.py — clean

• Golden masters untouched

Notes

• The Docs workflow builds the two new pages on merge; the MyST CLI is not installed here, so they were reviewed in source. Links point at existing targets (weco-rules.md, sds-detection.md, the derived-variables tutorial).

🤖 Generated with Claude Code

https://claude.ai/code/session_01BKphCi1LC1hjGYteWEK9JV