Keep the server error when the connection closes mid-operation

[RitiGrover]

When the server sends an ErrorResponse and then closes the connection without a ReadyForQuery, the error was dropped. The waiter only got a generic ConnectionDoesNotExistError. This happens with pgbouncer's query_wait_timeout and with any FATAL such as pg_terminate_backend().

_handle_waiter_on_connection_lost now checks for a pending failed result. The exception type stays ConnectionDoesNotExistError so existing handlers keep working. The server message is appended and the PostgresError is set as __cause__:

asyncpg.exceptions.ProtocolViolationError: query_wait_timeout

The above exception was the direct cause of the following exception:

asyncpg.exceptions.ConnectionDoesNotExistError: connection was closed in the middle of operation: query_wait_timeout

Tested with a new test that terminates a backend mid-query. It fails on master and passes with the fix. I also checked the pgbouncer case against a small fake server that sends FATAL 08P01 query_wait_timeout and closes. The full suite passes on PostgreSQL 15 apart from the GSSAPI tests, which need Kerberos binaries I didn't have locally.

Fixes #1299

[elprans]

LGTM, thanks!