Allow expand writtenLines + msPartsCount in expanded ODD

[duncdrum]

Summary

• Catch expanded ODD up to locked BetMasWeb expand-normalize encoding for range indexes:
  • empty <writtenLines quantity="…"/> under layout[@subtype='computed'] (replaces TEI design: computed physical dimensions (parallel @subtype siblings) #45/#44 docs that used layout/@writtenLines on the computed sibling)
  • trailing <msPartsCount quantity="…"/> on msDesc (materialized count(msPart); TEI design: computed physical dimensions (parallel @subtype siblings) #45 had left this as a live count)
• Authoring ODD unchanged. No change to citeStructure/@unit (token / no whitespace stays; multi-word units are an expand fix).

RNG regenerates on merge to master via existing Schema CI.

Failure taxonomy (full-corpus RNG ~8%)

Class	Whose job	Notes
writtenLines element under computed layout	Schema (this PR)	Expand intentional; current RNG only allows phrase content under layout
msPartsCount last child of msDesc	Schema (this PR)	Expand intentional; RNG expects only msPart/msFrag at end
citeStructure/@unit with spaces (page break, label prose)	Expand	teidata.enumerated correctly forbids whitespace; normalize units in expand
Leftover <source-url>	Expand	Non-TEI artefact; strip, do not schema-allow
Many sourceDesc / unexpected msDesc style reports	Often cascade	Illegal child (e.g. msPartsCount) makes parent alternatives fail

Do not re-enable expanded corpus RNG CI until this lands and expand cleans citeStructure units / source-url.

Refs: Schema #45, BetMasWeb expand-normalize-dimensions.

Test plan

• Schema CI green (ODD jing + TEIGarage RNG build)
• After merge: spot-check jing on files with computed layout + msPartsCount (e.g. UNESCO702, DabSey001) against regenerated tei-betamesaheft-expanded.rng
• Confirm files with unit="page break" still fail until expand fix

Made with Cursor