chore(mysql): apply existing _quote_ident to all SQL paths (#442 follow-up)

[yodakanohoshi]

Summary

MySQLDestination already has a _quote_ident helper that correctly handles schema-qualified table names (db.table → `db`.`table`), but it's only used in the swap path. The non-swap replace / insert / upsert paths still use raw f-string interpolation, so schema-qualified table names break in those modes.

Background

#442 / PR #498 closed the equivalent gap on the Postgres side (psycopg2.sql.Identifier with (schema, relation) parts). The MySQL audit was listed in #442's "Proposed Approach" but explicitly out of scope — this issue captures it as a follow-up.

The good news: unlike Postgres, the MySQL fix needs no new helpers. _quote_ident already exists in drt/destinations/mysql.py and does the right thing:

@staticmethod
def _quote_ident(table: str) -> str:
    if "." in table:
        return "`" + "`.`".join(table.split(".")) + "`"
    return f"`{table}`"

The work is purely mechanical: replace the remaining raw-backtick f-strings with _quote_ident().

Call sites that still use raw f-strings

Method	Line	Current	Needs
_load_replace	L190	f"TRUNCATE TABLE \{table}`"`	f"TRUNCATE TABLE {self._quote_ident(table)}"
_build_insert_sql	L353	f"INSERT INTO \{table}` ..."`	_quote_ident(table)
_build_upsert_sql	L368	f"INSERT INTO \{table}` ..."`	_quote_ident(table)
_build_upsert_sql	L372	f"INSERT IGNORE INTO \{table}` ..."`	_quote_ident(table)
get_row_count	L144-149	inline split+backtick logic	reuse _quote_ident

(_load_replace_swap and finalize_sync already use _quote_ident — leave them.)

Reference PR

PR #498 for the equivalent Postgres work is a near-perfect template — see especially the test patterns in tests/unit/test_postgres_destination.py::TestQualifiedIdentifiers and the test_*_uses_schema_qualified_identifier cases. The MySQL test file (tests/unit/test_mysql_destination.py) follows the same mock_connect pattern, so the diff should look very similar.

Acceptance Criteria

• All four _load_replace / _build_insert_sql / _build_upsert_sql call sites use _quote_ident
• get_row_count reuses _quote_ident instead of duplicating the split logic
• Unit tests cover schema-qualified table names (e.g. mydb.scores) for replace, upsert, and get_row_count paths — mirroring the cases added in PR fix: quote qualified Postgres table identifiers #498
• No new type: ignore; ruff check and mypy drt pass

Out of Scope

• Adding support for arbitrary quoted identifiers (embedded ., `, etc.) — not worth supporting until requested.
• ClickHouse audit — tracked separately.

Good First Issue

This is well-scoped for first-time contributors:

• No design decisions — the helper already exists
• Test scaffolding is already there; just add ~3 new test methods
• Reference PR (fix: quote qualified Postgres table identifiers #498) shows the exact pattern to follow