Task: Migrate ores.qt.refdata's history dialogs to the generic HistoryDialog

Retire the remaining hand-rolled per-entity *HistoryDialog classes in projects/ores.qt/refdata, migrating each entity's controller onto the generic HistoryDialog via full codegen regeneration (same process as AA97E310, which already migrated 26 of the ~43 original entities). 30 entities remain with a *HistoryDialog.cpp on disk; 14 of those have a codegen model and can be migrated the same way as AA97E310. The other 16 ( catalog, change_reason, change_reason_category, code_domain, coding_scheme, coding_scheme_authority_type, data_domain, dataset, dataset_bundle, floating_index_type, leg_type, methodology, nature_dimension, origin_dimension, payment_frequency_type, subject_area, treatment_dimension) have no codegen model at all — hand-crafted data-catalogue and change-reason-system infrastructure — and need a different, non-codegen migration approach or explicit scoping-out; not attempted in this pass.

• All 14 codegen-model-backed entities (asset_class_code, currency_market_tier, currency_pair_convention, day_count_fraction_type, ledger_feed_type, monetary_nature, party_type, purpose_type, regulatory_book_type, rounding_type, tenor_anchor, tenor_kind, tenor_resolution_algorithm, tenor_unit) migrated onto the generic HistoryDialog via full, unmodified codegen regeneration; legacy *HistoryDialog.{hpp,cpp,ui} files deleted per entity.
• Each entity's server-side history-provider registration added to registrar.cpp, following the AA97E310 pattern.
• No selective/hand-patched regeneration — full generator output per entity, diffs reviewed before applying, anything unexpected flagged rather than silently forced through or silently reverted.

(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.)

Same process as AA97E310: full, unmodified compass codegen entity generate <entity> --address ores.cpp.qt, diff reviewed before applying, legacy *HistoryDialog.{hpp,cpp,ui} deleted per entity, server-side history-provider registration added to registrar.cpp.

Of the 14 entities, 7 (asset_class_code, currency_market_tier, monetary_nature, rounding_type, tenor_kind, tenor_resolution_algorithm, tenor_unit) never had a history field mapper generated at all despite having a history protocol — generated via --address ores.cpp.presentation for all 7.

3 of those 7 (currency_market_tier, rounding_type, monetary_nature) turned out to be significantly stale beyond just the missing mapper: their domain object was missing tenant_id (the DB column and repository entity already had it — no schema migration needed), no pagination/count support, response structs missing the success=/=message failure-signalling wrapper (the exact silent-failure-as-empty-list bug fixed everywhere else in AA97E310), NATS subjects using dashes instead of underscores, and no event registrar. Flagged to the story owner before proceeding (full regen here is materially bigger than "migrate a history dialog"); decision was to do the full --address ores.cpp regen for all 3 and review each diff carefully — confirmed no .sql files in any of the three diffs, so no schema migration was required. Also caught two further stale delete_request_id_field=/=history_request_id_field overrides this surfaced (currency_market_tier: tier->=codes=/=code=, rounding_type: type->=codes=/=code=, monetary_nature: nature->=codes=/=code=) — same class of bug as AA97E310's 8 conventions, just discovered a step later since these entities' protocols were stale enough that the delete/history requests didn't even compile until the entity-wide field rename was applied. Also required a CMake reconfigure (new event-registrar .cpp files aren't picked up by an incremental build) and a hand-fix to ores.qt/api/src/LookupFetcher.cpp, whose hand-written fetch_rounding_types=/=fetch_monetary_natures=/ =fetch_currency_market_tiers referenced the old response field names.

Manual QA (scenario 44064405) caught one further real bug: saving a rounding_type record didn't fire a live-eventing notification. Root cause: refdata_rounding_types_notify_trigger_create.sql existed on disk (correctly codegen-generated) but was never \ir-included in create/refdata/refdata_create.sql — a pre-existing gap unrelated to this session's other changes, so no environment ever had the trigger. Applied the trigger directly to the running DB and fixed the missing include for future recreations. While investigating also found (and fixed, drive-by) 15 more refdata entities entirely missing from refdata_drop.sql's narrower per-table teardown list (their create side was fine, confirmed live in the DB) and captured the root cause — validate_schemas.sh discovers SQL files via a filesystem glob rather than walking the actual \ir chain, so a missing include is invisible to it — as its own backlog item, since it already does this reachability check for RLS policy files specifically and just needs generalizing.

Scope note: the remaining 17 entities under projects/ores.qt/refdata with a *HistoryDialog.cpp but no codegen model reachable from ores.refdata itself are tracked by the follow-on task Migrate remaining ores.qt.refdata entities without a codegen model onto the generic HistoryDialog, not this one. 5 of those 17 turned out to actually have a codegen model under a different owning component (ores.trading, ores.dq) once investigated properly; the other 12 are hand-crafted change-reason/coding-scheme/data-catalogue infrastructure needing a scoping decision before any migration.

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

14 codegen-model-backed entities migrated onto the generic HistoryDialog via full, unmodified codegen regeneration: asset_class_code, currency_market_tier, currency_pair_convention, day_count_fraction_type, ledger_feed_type, monetary_nature, party_type, purpose_type, regulatory_book_type, rounding_type, tenor_anchor, tenor_kind, tenor_resolution_algorithm, tenor_unit. Fixed 7 missing history field mappers, brought 3 significantly stale entities' full backend stack in line with codegen conventions (no schema migration needed), fixed 3 more stale delete_request_id_field=/=history_request_id_field overrides, and fixed a pre-existing, unrelated SQL gap (rounding_type's missing NOTIFY trigger include) that manual QA surfaced. QA scenario 44064405 PASSED after all fixes.

Remaining 17 non-codegen-backed entities tracked separately by Migrate remaining ores.qt.refdata entities without a codegen model onto the generic HistoryDialog.