Fix schema-sync indexes for Postgres schemas

[ouywm]

PR Info

• Closes schema-sync creates indexes against unqualified table names for entities in non-default PostgreSQL schemas #3084

New Features

• None

Bug Fixes

• Fix schema-sync generated indexes for entities in non-default PostgreSQL schemas.

  Schema-sync can already discover and match tables in non-default PostgreSQL schemas after entity sync does not work with postgres when tables have schemas defined #2952 / Fix schema sync not discovering tables in non-default schemas #3016. However, indexes generated from entity attributes still used the unqualified table name.

  For an entity declared with #[sea_orm(schema_name = "sys", table_name = "app_user")], generated index DDL could target ON "app_user" instead of ON "sys"."app_user". When the connection's search_path still resolves unqualified table names in public, sync can fail with relation "app_user" does not exist, or target a same-named table in the wrong schema.

Breaking Changes

• None. This is backward compatible. MySQL and SQLite index generation keep using unqualified table names.

Changes

• Use Entity::table_ref() for generated index targets on PostgreSQL, so schema_name is included when present.
• Keep MySQL and SQLite generated index targets unchanged.
• Add unit coverage for generated indexes on an entity with an explicit PostgreSQL schema.
• Add schema-sync regression coverage for #[sea_orm(indexed)] and multi-column #[sea_orm(unique_key = "...")] in a non-default PostgreSQL schema.
• Mirror the fix and tests in sea-orm-sync.