refactor(rds): extract engine defaults and final-snapshot helpers

[vieiralucas]

Summary

`create_db_instance` and `delete_db_instance` each had multi-line bodies that didn't need to live inline.

• `create_db_instance`: the port, parameter group, and logical database name defaults were all inline match-on-engine blocks. Move them into `default_port_for_engine`, `default_parameter_group` (which handles the `default.postgres16` / `default.mysql8.0` / `default.mariadb10.11` naming), and reuse the existing `default_db_name` helper. The operation body loses ~35 lines and reads top-to-bottom without stalling on default values.
• `delete_db_instance`: the 80-line 'create final snapshot' block (runtime check, lock-drop read of the source instance, `dump_database` call, write-locked snapshot insert) becomes its own `create_final_db_snapshot` method. The conflict check runs twice on purpose — once under the read lock to avoid paying for the dump, once under the write lock to make the commit atomic — and the helper preserves that.

No behavior change.

Test plan

• `cargo fmt`
• `cargo clippy --workspace --all-targets -- -D warnings`
• `cargo test -p fakecloud-rds` (14 passed)

---

Summary by cubic

Refactored RDS to extract engine defaults and final snapshot logic into helpers, making create_db_instance and delete_db_instance shorter and easier to read. No behavior change.

• Refactors
  • Moved default port and parameter group selection to default_port_for_engine and default_parameter_group; reused default_db_name.
  • Pulled final snapshot flow into create_final_db_snapshot, preserving the double conflict check and lock semantics.

^{Written for commit 76aaa05. Summary will update on new commits.}

[cubic-dev-ai]

No issues found across 1 file