[ronaldtse]

Summary

Adds two work orders to TODO.sota:

• 89-remaining-encodings-sota.md — the master list of every unencoded system
  (72 Track A items across BGN/PCGN, ALA-LC, DIN, BSI, ICS, ISO scope, ICAO/UNGEGN;
  63 D, 8 D+M, 1 M-part, 0 blocked), the Track B SOTA matrix (six model lines with
  current numbers and targets), and the graph path from reference document to map,
  tests, parity, silver/gold data, training, evaluation, quantization, release, and
  runtime.
• 90-wo90-bgnpcgn-cyrillic-batch.md — the P0 batch work order and its result:
  eight BGN/PCGN Cyrillic systems encoded (maps PR Add eight BGN/PCGN Cyrillic romanization systems maps#215, 777 tests
  passing, corpus strict-parse 300/300) and the Kurdish "2012 revision" resolved as
  a reissue of the 2007 system already in the corpus.