fix(tasks): Add autoretry for too many simultaneous queries error in some tasks

[webjunkie]

Problem

We throw a lot of simultaneous queries errors. We can retry.

Fixes POSTHOG-DNR
Fixes POSTHOG-DNR
Fixes POSTHOG-TMV

Changes

• use Celery autoretry_for
• but for that to work, isolate the exception as a real one

Does this work well for both Cloud and self-hosted?

n/a

How did you test this code?

• tested exception handling, added test
• not sure how to test locally for real cause needs reproducing too many queries somehow

[Twixes]

Looks good to me. More load if there's a period of lots of queries, but fewer user-facing errors for intermittent troubles, which seems like an alright treade-off

[Twixes]

Would also be good to move EstimatedQueryExecutionTimeTooLong and into this QuerySizeExceeded, in a separate PR