Initial working Rust binding connection pool

[dave42w]

Description

This adds a ConnectionPool instance to the rust binding Database. The default is for it to not be enabled, so same behaviour as default.

Motivation and context

The goal is to save execution time with a transparent ConnectionPool

Description of AI Usage

No AI used

[turso-bot]

Please review @jussisaurio

[dave42w]

If the ConnectionPool is turned on then one test fails:

failures:

---- tests::test_database_persistence_many_frames stdout ----

thread 'tests::test_database_persistence_many_frames' panicked at bindings/rust/src/lib.rs:525:22:
Query succeeded after WAL deletion and DB reopen, but was expected to fail because the table definition should have been in the WAL.
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    tests::test_database_persistence_many_frames

test result: FAILED. 11 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.06s

I think the issue is with the way Connection drop "recycles" Connections into the ConenctionPool. As drop cannot (yet) be async, we can't call self.maybe_handle_dangling_tx().await?;

I'd love some feedback on how to resolve this (maybe something better than cloning the inner Connection in drop?)