rethink postgres_exporter/common/pg16/queries_pg_stat_statements.yml

[hunleyd]

Describe the enhancement you are proposing:
(from #367 (comment))

This is more a critique of this query, and if this is already basically the same as in other pg versions I'll let it slide, but seems like this isn't necessarily a sensible metric, since it's not the mean of the weighted average of queries; I'd assume you'd really need to scale each mean_exec_time by the number of calls (or use total_exec_time again) and then divide by the sum of the number of calls total) to get the true value here. Something to think about, and maybe what we have here is "good enough" for most use cases (really seems like only very skewed query workloads would be drastically affected here).