Task: Use oresmd on the market data generation path

No runtime reader, feed or classifier composes an ORE key; the runtime identity is the oresmd URI, and ORE keys live only at the ORE XML boundary.

• The synthetic IR feed builds native oresmd datums and publishes the same series the bootstrap reads.
• The series classifier keys its rules on the oresmd identity, with the rule catalogue migrated.
• The oresmd codec accepts a valid datum that has no ORE spelling; ORE import and export keep their keys.

(Implementation strategy. Written when work starts; key decisions are distilled into the parent story's * Decisions at close, but the plan itself stays — it is the historical record of what we did.)

Every number this task claims about a tree, with the command that produced it and the commit it was measured at. A number without both is not a measurement, because a clean-up record describes a tree and trunk moves under a branch: a consumer count, a gate's component count and a suite's size all change with no code changing, and the reader of a bare number cannot tell which tree it came from. Where the raw output matters, put it in an evidence/ file beside the task and name that file in the Command column.

Manual QA scenarios (scaffolded via compass add test_scenario) that verify this task. Link new ones here as they're created; the scenario doc itself links back via its "Verifies task" field.

The runtime no longer spells market data the ORE way; the ORE key lives only at the ORE XML boundary.

• The synthetic IR feed builds each quote as a market data model datum through one builder (pillar_quote_datum); pillar_quote_key is deleted.
• The series classifier keys on (key_kind, instrument, quote); index series read their class from their own schema row; the catalogue migrates from 42 ORE rows to 41 datum rows and 1 index row, and the GENERIC-MD wrapper row is gone.
• The oresmd codec accepts every datum the model admits; the ORE-key validation is removed in read and write; the ORE import and export keep their keys.
• The seeded blueprint is the spot-starting USD SOFR ladder reading one named series, because a meeting-dated ladder's series identity depends on the as-of.

Evidence: marketdata core 330, service 203, synthetic 94, shell and refdata suites, web 2327, BFF 1137, codegen idempotent, model and docs checks clean. Live: the IR feed publishes 12-point batches, no stored series or observation carries an ORE spelling, and the migrated catalogue classifies the stored series.

Open: no synthetic-provider tick reaches the store in this environment, so the compute, check and publish are proved by the suites and not live; the smoke aborts at trade generation on a missing handler. The delivery plumbing is the next task.