[MarkParker5]

Log line (pairing a Tuya TS0001 relay):
ERROR majordom_zigbee.controller:controller.py:976 [PAIR] read_attributes error device=…(TS0001) endpoint=1(ON_OFF_LIGHT) cluster=1794(Metering) error=TimeoutError attrs=
plus, once on the CM4 run: ERROR asyncio:base_events.py:1871 Task exception was never retrieved.

Cause

• _read_cluster_attributes logged every exception at ERROR and kept reading the next chunks, so a listed-but-unanswered cluster (Metering on the TS0001) logged an ERROR per chunk and waited the 2 s timeout for each.
• _create_task (and the reconnect task) only discarded finished tasks; nobody retrieved their exceptions.

Fix

• A TimeoutError now logs once per cluster read and stops the remaining chunks: WARNING for optional clusters, ERROR still for main-parameter clusters (OnOff, LevelControl, …). Other exceptions stay ERROR. Timeouts unchanged.
• A _log_task_exception done-callback on every _create_task task and on the reconnect task logs the exception (ERROR with traceback) and so retrieves it.

Tested

• New tests (failed before the fix): test_an_unanswered_optional_cluster_read_is_a_warning_not_an_error (TS0001 with Metering timing out: only WARNINGs, one request per read), test_a_failed_background_task_is_logged_not_left_unretrieved.
• Full suite: 93 passed, 1 skipped. ruff check/format and ty clean.