fix: quote qualified Postgres table identifiers

[Photon101]

Closes #442.

Summary

• add helpers that split optional schema.table names and compose psycopg2.sql.Identifier with separate schema/relation parts
• use the helpers across Postgres row count, replace, swap, finalize, insert, and upsert SQL paths
• keep swap shadow/old suffixes on the relation name so marketing.email_events becomes marketing.email_events__drt_swap, not a single quoted identifier
• add regression coverage for schema-qualified upsert, replace, swap, row-count, and replace error paths

Validation

• uv run --extra dev --extra postgres python -m pytest tests/unit/test_postgres_destination.py tests/unit/test_dry_run_row_count.py tests/unit/test_dry_run_summary.py -v
• uv run --extra dev --extra postgres python -m pytest --cov=drt.destinations.postgres --cov-report=term-missing tests/unit/test_postgres_destination.py
• uv run --extra dev --extra postgres ruff check drt/destinations/postgres.py tests/unit/test_postgres_destination.py
• uv run --extra dev --extra postgres mypy drt/destinations/postgres.py
• git diff --check

[codecov]

Codecov Report

❌ Patch coverage is 46.15385% with 14 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
drt/destinations/postgres.py	46.15%	14 Missing ⚠️

📢 Thoughts on this report? Let us know!

[Photon101]

Updated in c83da81 after the initial Codecov report. I added coverage for the schema-qualified helper branches and replace-mode error paths; the latest codecov/patch status check is now green.

[yodakanohoshi]

Thanks for the follow-up on this one, @Photon101! 🙌

This is the right finishing touch for #442 — the helper trio
(_split_qualified / _qualified_ident / _with_relation_suffix)
cleans up every call site, and you caught the subtle bit that
ALTER TABLE ... RENAME TO only accepts a bare relation on the RHS.
Test coverage across upsert / replace / swap / row_count is exactly
what was asked for.

I'll add a CHANGELOG entry post-merge — no action needed from your side.

The MySQL / ClickHouse audit listed in #442's "Proposed Approach" is
out of scope for this PR and works well as a separate follow-up issue.
If you'd be interested in tackling it, I'd be glad to open one with
the scope written up — entirely no pressure, this PR stands on its own.

Thanks again for the contribution!