Interface for retries?

[MichaelChirico]

I wonder if there's been any thought put into a centralized interface for retrying queries.

It's quite common in distributed systems to hit timeouts, cache misses, etc., when running queries at scale, and we often observe users custom-rolling a retry interface to make their pipelines (etc) more robust to these temporary misses.

I think it would be useful to host such an interface centrally in DBI, perhaps directly in the dbSendQuery() generic? What might such an interface look like generically, a failure callback?