Task: Replace the asset-specific ticks with one market tick

Every live market data feed, whatever its asset class, publishes and ingests one tick type keyed by the oresmd quote URI, and every feed takes its owner from one source.

• fx_spot_tick and ir_curve_tick are deleted; one modelled market_tick carries the oresmd quote URI, the value, the observation time and the source.
• Every feed takes tenant, workspace and party from its feed binding; no tick carries an owner.
• The ingest loop reads the URI with oresmd_uri_codec and classifies the datum with classification_key_of; it no longer rebuilds an ORE key from parts or reads a classification off the wire.
• P02: marketdata.v1.tick is declared once, by the market_tick model, and every use of it, the republish subject builder included, references the generated constant.
• The republished subject is marketdata.v1.tick.<tenant>.<workspace>.<party>.<key>, where <key> is the canonical ORE key of the tick's datum, lower-cased with its slashes turned to dots, written by one builder that both the ingest loop and the client use.
• A tick finds its consumers by source through the enabled feed bindings of that source, for every feed; a tick from a source with no enabled binding is dropped with one warning per source.
• market_tick.value is decimal text, as stored observations are.
• Tenant provisioning creates bindings for the IR curve generation configs as it does for the FX spot ones.
• The synthetic FX and IR curve feeds publish market_tick, and a live feed run stores observations for both.

The wire changes producers and consumers together, so this is one pull request of commits that each build:

1. The tick. Model market_tick in ores.marketdata.operations, with :subject: marketdata.v1.tick, and write the one republished-subject builder in the api against its generated constant.
2. The binding. A binding maps a source to the tenant, workspace and party that consume it. Its oresmd_uri and asset_class go: the tick names its own datum, and a curve source publishes many datums across several series. Its natural key becomes the party and the source.
3. The ingest loop. One path for every tick: find the source's enabled bindings, read the URI with oresmd_uri_codec, classify with classification_key_of, store the observation, republish under the builder's subject.
4. The producers. The synthetic FX and IR curve feeds publish market_tick. Tenant provisioning creates a binding for every feed, FX and IR alike.
5. The deletions. fx_spot_tick, ir_curve_tick, their JSON I/O, the two copies of the subject function, and the FX-only client subscription, which becomes a subscription to market_tick.
6. Proof. The fleet runs both synthetic feeds, and observations from both arrive under their datum URIs.

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.

A live tick is one type for every feed and asset class: market_tick, the oresmd quote URI, the value as decimal text, the observation time and the source. Every feed takes its owners from its feed binding, which now maps a source to a party and nothing else, and the ingest loop has one path that classifies a new series from its datum, as the file import does. P02 holds for marketdata: marketdata.v1.tick is declared once, by the model, and the republished subject and both stream wildcards are built from it. On a recreated database, 90 of 90 suites passed, and a live ACME run stored FX and IR curve ticks from 33 of 34 sources; the one that published nothing fails on a pre-existing pillar defect, recorded as a capture.