Rewrite merge methods on real Core constructs so schema_translate_map applies to loaded data

[nicoloesch]

Summary

Depends on AustralianCancerDataNetwork/oa-configurator#37.

Requires pyproject.toml version update for oa-configurator

Breaking change

OrmLoaderConfig.test_orm_db is gone, replaced by test_orm_db_pg/test_orm_db_sqlite. Anyone with test_orm_db set in their omop.config will need to update it. There is no back-compat shim. Needed because dialect-specific integration tests require a guaranteed-correct-dialect test database each, which one shared field couldn't express.

Root cause

Four related ways schema_translate_map was silently bypassed across orm-loader:

1. merge_replace/merge_insert/merge_upsert built raw, unqualified text() SQL
2. staging-table index inspection used a schema-blind sa.inspect()
3. materialized-view lifecycle methods took a hand-threaded schema string instead of resolving through the connection, and
4. a staging-table reflection path could open a second connection that, on SQLite, silently discarded a session's own uncommitted work.

Fix

Loaded/merged data

• merge_replace/merge_insert/merge_upsert rewritten against real Table objects and
  dialect-specific Core constructs (sa.delete, insert().from_select(),
  postgresql.insert().on_conflict_do_nothing(), with a SQLite-specific counterpart), for both backends.
• Fixed the one remaining raw isolation_level="AUTOCOMMIT" call site.
• Staging-table reflection reflects now through the session's ownconnection() to prevent data loss

Materialized views

• create_mv/refresh_mv/drop_mv/index creation are fully role-aware
  • each resolves its physical schema from the bound connection's own schema_translate_map (__mv_schema_tag__, defaulting to Role.PRIMARY)
• The Postgres-only dialect guard is now a single check inside the shared _as_connection() context manager, replacing a narrower per-method check that only ever guarded direct/manual PostgresBackend() construction.
• Staging-table index management (manage_indices) now resolves its Inspector schema-aware, via physical_schema_of(session, schema_tag=validate_schema_tag(cls.__table__)).

Split primary/vocab connections

• Loading and merging now works correctly against genuinely separate primary/vocab connections

Config, CI, test infrastructure

• STAGING_SCHEMA's reserved-schema registration moved into config.py (the actual omop.config entry point), so it's guaranteed to run whenever this package's config is resolvable
• OrmLoaderConfig.test_orm_db split into test_orm_db_pg/test_orm_db_sqlite for dialect-specific test configurations
• CI split into separate SQLite and Postgres build-test jobs
• engine_with_replica_role now uses autocommit_connection()'s Engine branch directly
• Removed own Dialect system in favour of the unified oa-configurator one
• Adopts oa-configurator's new test infrastructure (isolated_test_database, DIALECT_PARAMS, db_dialect marker).
• New tests:
  -test_schema_translate_map.py: the actual regression test for the originally-reported
  bug,
  • test_split_connection.py, test_shared_backend.py: merge-method contract tests unified across both dialects,
  • test_reserved_schema.py: proves STAGING_SCHEMA registration is actually
    enforced cross-package

Schema drift protection, primitive consolidation

• helpers/bootstrap.py::create_db/bootstrap() is now wrapped in guard_schema_provenance_for via oa-configurator's open_connection to prevent schema drift
• helpers/sql.py::qualify_identifier deleted entirely
  • consolidated into oa-configurator's qualified()
• backends/base.py::_as_connection now delegates its Engine/Connection branch to oa-configurator's open_connection.
• Adopts physical_schema_of stack-wide
  • fixes validate_schema_tag usage in materialised_view_mixin.py and schema_if_supported usage in backends/sqlite.py.

Checklist

• Applied exactly one label (breaking, feature, fix, dependencies, or chore)
• Tests pass locally (uv run pytest -q): 246 passed, 1 xfailed
• Tests pass locally (uv run pytest -q -m postgresql): 37 passed
• Lint passes (uv run ruff check .)

[gkennos]

raise dependency base for oa-configurator

also depending how you resolve the comment that remains here, you may need to update oa-configurator --> sql.py as well (validate_schema_tag)

[gkennos]

manage_indices now defers schema resolution until there are indexes to rebuild, which fixes the crash under index_strategy="keep", so that's good, but postgres still defaults to drop_rebuild so a legit literal schema still calls validate_schema_tag, which raises for anything that isn't a registered Role

Please add a third case that preserves any unrecognised string as a literal physical schema, or explicitly narrow/version the public contract to document that literal schemas are no longer supported.