Add a provider-level query timeout mechanism

[data-sync-user]

Given a search query, Merino will query it against all the providers (either the default providers or the specified ones). To reduce the query duration, all those query requests are executed concurrently. Merino will wait until all of the providers respond. That behavior may not be ideal for providers served by remote services as Merino could wait indefinitely if the service outage takes place.

Thus, we should add a timeout mechanism to Merino so that it will stop waiting once the timeout reaches; respond to the request with the already returned suggestions; possibly cancel the remaining requests and log the timeout event/metric.

┆Issue is synchronized with this Jira Task
┆Epic: Dynamic Wikipedia results on Firefox Suggest

[data-sync-user]

➤ Nan Jiang commented:

Notably, this can be readily implemented by the new asyncio.timeout() ( https://docs.python.org/3.11/library/asyncio-task.html#asyncio.timeout ) introduced in Python 3.11.

[data-sync-user]

➤ Nan Jiang commented:

#112 ( https://github.com/mozilla-services/merino-py/pull/112|smart-link )

[data-sync-user]

➤ Nan Jiang commented:

Merged.