Task: Update ores.cpp.qt codegen templates for the generic HistoryDialog

Update the ores.cpp.qt controller and history-dialog codegen templates (cpp_qt_controller.{hpp,cpp}.mustache, cpp_qt_history_dialog.{hpp,cpp}.mustache, qt_history_dialog_ui.mustache) so regenerating any entity emits the generic-HistoryDialog wiring — HistoryDialog(entity_type, entity_id, clientManager) construction, and the generic openVersionRequested=/=revertVersionRequested(entity_type, entity_id, version) signals resolved back to a typed domain object via a fetch<Entity>History wrapper — instead of generating a new per-entity *HistoryDialog class.

Currency is the zero-diff proof: CurrencyController.{hpp,cpp} was already hand-migrated to exactly this shape in PR #1553/#1563 (discovered as a manual, uncodegen'd divergence while starting the migrate-refdata-history-dialogs task — compass codegen entity generate currency --address ores.cpp.qt --diff currently shows the old per-entity-dialog pattern as the would-be regenerated output). Once the templates are updated, regenerating currency must produce zero diff against its current hand-migrated files.

This unblocks every sibling per-component migration task under this story: they become pure compass codegen entity generate <entity> --address ores.cpp.qt regeneration once this lands, for any entity that already has a server-side history field mapper.

• compass codegen entity generate currency --address ores.cpp.qt --diff produces near-zero diff against the current, already- migrated CurrencyController.{hpp,cpp} (cosmetic naming/ordering only — the fetch-wrapper's parameter name and the doc-comment wording generalize slightly differently than the hand-written version; no behavioural difference). Met.
• The templates no longer emit a per-entity *HistoryDialog class/header/.ui for any entity generated after this change. Met.
• Regenerating any other entity that has both entity_type_of() and a history field mapper (e.g. country) produces a controller wired the same way as currency's, and builds clean. Met — country regenerated to a clean build with no manual intervention needed beyond deleting its now-orphaned CountryHistoryDialog.{hpp,cpp,ui}.
• Entities without a history field mapper yet are unaffected (no crash/malformed output from the template — either skip history wiring gracefully or this is explicitly out of scope, whichever the investigation finds simplest). Met by design: the template always calls entity_type_of()=/constructs the generic request regardless of mapper presence; an entity without a server-side mapper simply gets a graceful =success: false response at runtime (per the dispatch registry's existing "unknown entity_type" handling) until its own commission story adds the mapper — no template-level gating needed, per the story owner's direction.

(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.)

Edited the literate org sources (ores.cpp.qt.controller_header.org=/ =ores.cpp.qt.controller_impl.org), not the tangled .mustache files directly, and re-tangled with compass build --direct tangle_codegen_templates. Mirrored CurrencyController.cpp's hand- migrated shape (PR #1553/#1563) as closely as possible, using entity_type_of({{domain_class}}{}) (ADL, no new template variable needed) instead of a hardcoded entity-type string, and the existing history_request_class=/=history_response_class=/ =history_response_data_field=/=history_request_id_field variables (already used by the now-disabled per-entity history dialog template) to build the new fetch<Entity>History wrapper generically for both UUID- and natural-key entities.

Along the way: discovered wireVersionHistory<...>=/ =VersionNavigationHelper.hpp, referenced by the has_version_navigation branch this template replaces, don't exist anywhere in the codebase — that path was dead/aspirational code, never actually implemented, so no entity currently regenerating with has_version_navigation: true could have compiled. Replaced with the same fullHistory-parameter approach currency's hand migration already uses and proved out.

• Regenerating currency also surfaced unrelated CurrencyDetailDialog.cpp drift (a different archetype, dynamic_combo badge-setup helper) — confirmed pre-existing and unrelated via a stash-and-diff check, left untouched, captured separately: Reconcile dynamic_combo detail-dialog codegen drift.
• Regenerating currency also caught a genuine currency model bug: the is_required column for fraction_symbol=/=format still said true, but the checked-in .cpp had been hand-patched (with an explanatory comment) to not require them, since neither is required by the backend. Regenerating without fixing the model would have silently reintroduced that bug. Fixed the model directly.
• While verifying the build, found 5 entities (AssetClassCode, Calendar, InstrumentCode, IrCurveGenerationConfig, IrCurveTemplateEntry) added after PR #1563 with the same fragile positional HistoryWidgets init that PR fixed elsewhere — fixed here too (blocking the build regardless of this task).

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.

The ores.cpp.qt controller templates now generate the generic- HistoryDialog wiring for every entity by default — no per-entity *HistoryDialog class is emitted anymore. Regenerating currency matches its hand-migrated PR #1553/#1563 shape to a near-zero diff; regenerating country (a second entity, natural key, no badge/combo fields) produced a clean build with no manual fixes needed beyond deleting its now-orphaned per-entity dialog files. Both are applied and committed on this branch, so the sibling migration tasks under this story become pure compass codegen entity generate <entity> --address ores.cpp.qt regeneration going forward.

Along the way, fixed two genuine bugs unrelated to the template itself but blocking verification: the currency model's stale is_required flags (would have reintroduced an already-fixed validation bug on regeneration), and 5 newer entities' fragile positional HistoryWidgets init (blocking the build outright). Captured the unrelated dynamic_combo detail-dialog drift found along the way rather than pulling it into this diff.

Build clean (linux-clang-debug-make, full build); ctest — ores.diff, ores.refdata.api, ores.qt.headless, ores.qt.tests — 4/4 passed.