After a pg_lakehouse error, subsequent queries crash with "there is no reactor running"

[neilyio]

Bug Description
After a runtime error in a pg_lakehouse query, we're stuck with:

Query 1 ERROR at Line 1: : ERROR:  there is no reactor running, must be called from the context of a Tokio 1.x runtime

How To Reproduce
Follow quickstart to CREATE FOREIGN TABLE trips.

Turn off wifi, so that the extension cannot talk to AWS.

Run SELECT COUNT(*) FROM trips;, which should produce:

   error sending request for url (https://s3.us-east-1.amazonaws.com/paradedb-benchmarks/yellow_tripdata_2024-01.parquet): client error (Connect): dns error: failed to lookup address information: nodename nor servname provided, or not known: failed to lookup address information: nodename nor servname provided, or not known

Run SELECT COUNT(*) FROM trips; again, which produces:

Query 1 ERROR at Line 1: : ERROR:  there is no reactor running, must be called from the context of a Tokio 1.x runtime

The error persists until the extension is reloaded.