Merge Liber Usualis local-deployment bug fixes (#981) before production indexing

#981 fixes a real bug (#979) where MEIParser silently drops every staff but the first on multi-staff pages -- 67% of the Liber Usualis corpus's 2318 pages have 2+ staves (up to 16 on one page). Without this fix, production indexing of the Liber Usualis would silently produce incomplete OMR data, the same way it did locally before the fix.

This needs to land (and any review feedback addressed) before real production indexing work in this migration (see the "Run production OMR indexing" sub-issue) can produce correct results.