Prometheus feature `native-histograms` breaks targets with `info` metric types from being scraped.

[defenestration]

I was updating fluxv2 and using this repo as a template, but i noticed in prometheus, that the kube-state-metrics target was showing the error "invalid metric type "info". After some some local testing with a fork of the repo, I'm finding the native-histograms feature of prometheus seems to cause the problem. I have a branch here for testing this out.

https://github.com/defenestration/flux2-monitoring-example/tree/native-histograms

Heres a pr with the diffs from this repo from when i forked it if it helps https://github.com/defenestration/flux2-monitoring-example/pull/1/files

Realize this is probably an upstream issue, but wondering what the best work-around for this could be? Seems like the custom metric types may be best if set as Gauge instead perhaps?

I will continue to look on my own but figured it was worth filing an issue in case someone else had this problem too.