Note that this complicates the APIs, since it pushes the responsibility
of declaring a query a read or write to the user to then be scheduled
appropriately.
Add `.(read_|)query_row_f` APIs similar to rusqlites
`conn.query_row` accepting a `|row| -> Result<T>` to reduce the use of
`Row` and `Rows`.
Make benchmarks more isolated by not sharing a DB across runs
accumulating writes.
These implementations are dangerous since despite being behind an async
API, they block on I/O and can thus clog up the tokio runtime. Which is
fine if SQLite is all you do but really bad if other, independent work
gets stalled.
