docs(xlsx): confirm line-legend marker floor rule, document blockers (#1618)

[developer0hye]

Summary

• Confirms native Excel's line-legend marker placement rule for XLSX: line-legend circles omit native placement within the sample #1618 — centre = floor(target) - (marker_size mod 2)/2, the same algebra WorksheetMarkerPlacement already applies to plot markers — against 9 legend sizes across two independent native Excel export batches (4 gathered fresh for this PR, specifically to get the sample line's own x-extent, which the earlier XLSX: small-font legends retain oversized line-series markers #1617 batch did not record), on both axes, including the rows that discriminate floor from round.
• Documents precisely why a general code fix cannot land yet: the x axis is algebraically reachable but boundary-sensitive under the Segoe UI substitute this machine renders (floor margins as small as 0.053pt); the y axis is blocked by two separate issues — office2pdf's own sample-line position already drifts up to 2.87pt from native past 9pt (a distinct root cause, filed separately as XLSX: line-legend sample line doesn't scale with legend font size past the marker cap #1753, since this project's convention is one issue per root cause), and even once that's fixed, Rust doesn't currently have the sample line's resolved absolute position to floor (it comes out of Typst's own inline box/baseline layout).
• No runtime code changes — assets/validation/issue-1618/ only, plus a new issue (XLSX: line-legend sample line doesn't scale with legend font size past the marker cap #1753) for the unrelated root cause discovered during this investigation.

Related issue

Related: #1618, #1617, #1753

Testing

• python3 assets/validation/issue-1618/verify_floor_phase_rule.py exits 0, confirming the floor rule against all 32 rows of both native datasets.
• python3 assets/validation/issue-1618/measure_legend_line_x.py 0.82 <pdf>... reproduces the x0/x1/mid values committed in native-legend-line-key-measurements.json.
• Documentation-freshness review completed (one inaccurate cross-dataset claim found and fixed before this commit).
• No .rs/runtime config changes — tests skipped per project convention for non-runtime changes.

Visual impact

• No rendered PDF change
• Rendered PDF change or visual evidence added
• Reason: Adds only assets/validation/issue-1618/ measurement scripts and native-export data used to derive and verify the placement rule; no chart rendering or codegen path is touched, so there is no rendered PDF change to audit.

Checklist

• Commits include a Signed-off-by line
• PR scope contains one root cause
• Remaining converter or harness deviations each reference an open issue

🤖 Generated with Claude Code