This might get fixed by T347710

I'm not 100% sure that this ticket is necessary any more. For context, it was created at a time when we had only 5 presto workers and our users were experiencing frequent failures to render graphs using the Presto/Hive backend in Superset.
We made many adjustments to try to alleviate the issue, mostly by increasing the timeout period: e.g T299141, T294768, and T294771.

We also created some documentation around the problem: T294046: Write document about "Fast Enough Superset" and the resulting document.

The most notable improvement came about as a result of tripling the size of the presto cluster: T323783: Add an-presto10[06-15] to the presto cluster
However, adding more presto workers initially made the performance problems worse it took several more tickets to find the cause of the performance issue: T325809: Presto is unstable with more than 5 worker nodes.

We eventually found that the problem was caused by the large flood of kerberos requests associated with the intra-cluster presto authentication: T329831: KDC performance tuning for TCP requests

Since increasing the number of kerberos KDC worker daemons, this has stablised the Presto cluster performance and we have not had any recent reports of superset queries timing out.
We have a ticket about creating a general purpose presto query logger: T269832: Add a presto query logger which is likely to provide a more comprehensive mechanism for gathering data about particularly slow or problematic presto queries.

If we do wish to go ahead with this ticket as it stands, then we might want to look into the custom statsd logging options, as explained here: https://superset.apache.org/docs/installation/event-logging/#statsd-logging

However, I'm not yet convinced of the value of it compared with some of our other work, given that our users are not complaining about Superset performance at the moment. @Gehel what do you think? Should we decline the ticket?