set_query_timeout() blocks while a query runs and stalls every Python thread #63

This work was produced with the help of language models.

The engine applies the query timeout (and set_max_threads_for_exec()) under the connection lock that a running query holds for its whole duration. Connection.set_query_timeout() calls the engine right away, so with a query in flight it blocks until that query finishes, and the pybind bindings wait holding the GIL, so every Python thread stalls meanwhile. AsyncConnection.set_query_timeout() with 4 long queries in flight froze the event loop for 12-13 s (0.21.2, CPython 3.13/3.14).

import threading, time, ladybug as lb

conn = lb.Connection(lb.Database(":memory:"))
q = "WITH range(1, 30000) AS r UNWIND r AS x UNWIND r AS y RETURN sum(x * y)"
threading.Thread(target=conn.execute, args=(q,)).start()
time.sleep(0.3)
started = time.monotonic()
conn.set_query_timeout(60_000)
print(time.monotonic() - started)

Expected: set_query_timeout() returns at once, and other threads keep running.
Actual: it returns when the query finishes, seconds later, and no other Python thread runs meanwhile.

Reproduced with the ladybug==0.21.2 wheel on CPython 3.13.16 and 3.14.8, and on main (548c183) with CPython 3.12.

A per-call execute(..., timeout_in_ms=...) is proposed separately in (to follow).

Fix: #67