Use prompt_toolkit's complete_in_thread option

[rolandwalker]

Description

This makes the interface more responsive by not blocking typing.

This is intended to combine with #1547 for more fluid typing.

Checklist

• I added this contribution to the changelog.md file.
• I added my name to the AUTHORS file (or it's already there).
• To lint and format the code, I ran

  uv run ruff check && uv run ruff format && uv run mypy --install-types .

[scottnemes]

Nice, seems a little more snappy

[rolandwalker]

@scottnemes the difference is most noticeable on a very long statement, since we are parsing the SQL to produce the completions.

[github-actions]

Review summary: Small change, but it alters threading behavior in completions. I see two correctness risks that should be addressed before merge.

1. Potential incompatibility with minimum prompt_toolkit version (hard failure).
   PromptSession will raise TypeError if the installed prompt_toolkit does not support complete_in_thread. The minimum version is prompt_toolkit>=3.0.6, but it’s unclear if complete_in_thread exists that far back. If it was introduced later, this PR breaks users pinned near the minimum.
   Action: confirm the option exists in 3.0.6. If not, either bump the minimum version or gate the argument based on prompt_toolkit.__version__.
   Refs: mycli/main.py:1184, pyproject.toml:15

2. Possible data race between background completion and completer mutation.
   With complete_in_thread=True, SQLCompleter.get_completions() runs on a background thread. But the completer is mutated/reset under a lock (reset_completions()), and the prompt toolkit thread does not acquire that lock. This can lead to inconsistent reads or transient errors when reset_completions() runs during a completion request.
   Action: wrap the completer with a thread-safe adapter (e.g., a custom Completer that acquires _completer_lock around get_completions), or ensure reset/swap are done in a thread-safe, immutable fashion.
   Refs: mycli/main.py:1181-1188, mycli/main.py:1388-1416, mycli/sqlcompleter.py:953-972

Tests: I didn’t run tests. There’s no test coverage for threaded completion behavior; consider adding at least a concurrency smoke test around reset_completions() + get_completions() if you keep background completion.

If you want, I can draft the locking wrapper or the version-guard patch.