Reduce "start-up" time for tasks in CeleryExecutor

[ashb]

This is similar to #11327, but for Celery this time.

The impact is not quite as pronounced here (for simple dags at least)
but takes the average queued to start delay from 1.5s to 0.4s

Closes #6905 - the config option added for LocalExecutor is used here too.

Data on this for a simple 10-task sequential DAG:

SELECT execution_date,
    min(start_date - queued_dttm) AS min_quued_delay,
    max(start_date - queued_dttm) AS max_queued_delay,
    avg(start_date - queued_dttm) AS avg
FROM task_instance
WHERE dag_id = 'scenario1_case2_03_1' GROUP BY execution_date;

execution_date	min_quued_delay	max_queued_delay	avg	with change?
2020-10-08 01:00:00+01	00:00:00.348837	00:00:00.473693	00:00:00.396751	Yes
2020-10-08 02:00:00+01	00:00:01.432304	00:00:01.574801	00:00:01.478422	No

This was discovered in my general benchmarking and profiling of the scheduler for AIP-15, but it's not tied to any of that work. There are more of these kind of improvements coming, each unrelated but all add up.

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.

[turbaszek]

Do we need to fork it? Shouldn't we just execute it in current process (celery worker process)?

[ashb]

Can't cos of the logging.shutdown() at the end of task_run (which we need to keep, as that's when remote logs are uploaded. #11327 (comment)