test: cover enum and array schemas and migrated-schema execution

[omer-cengel]

Summary

Adds acceptance tests showing that PostgreSQL schemas shaped like published sqlc examples load as documented, and that a migration directory runs end to end: PostgreSQL applies the files, and code generated from the same directory executes against the migrated tables.

Changes

• Schema-parser test: a single-file schema with an enum type, an enum column, a varchar[] column, indexes, and a dollar-quoted function followed by another statement loads, with the enum and array columns recorded as unsupported types.
• Schema-parser test: a migration file containing COMMENT ON TYPE fails on that statement with a located syntax diagnostic (Encountered unexpected token: "TYPE" at line 14, column 12).
• Schema-parser test: without that statement, the migration files applied in order ignore the COMMENT ON TABLE/COLUMN statements and the enum type, rename a table, drop one column and add another, and record the enum, enum-array, and text[] columns as unsupported.
• PostgreSQL integration test: migration files written to a temporary directory are applied in file-name order, and a repository generated and compiled from that directory creates and reads rows through the renamed table and its added column.
• The PostgreSQL test helper can now generate from a schema directory as well as a single schema file, and database reset drops the new fixture tables and enum type.

Scope and non-goals

• Test-only change: no compiler, runtime, configuration, or documentation behavior changes.
• Fixtures are original SQL; no third-party example files are added to the repository.
• Enum and array types stay unsupported, and COMMENT ON TYPE stays a syntax error; sqlcj does not pre-process SQL to skip it.