Chore: refactor repo filters to use more performant histogram queries

[jpmcb]

Also wondering if we can use these types of histograms for the repo filters on the explore page

Originally posted by @brandonroberts in https://github.com/open-sauced/api/pull/593#discussion_r1500686637

---

The queries used in discovering the repos in the repos filter are not very performant: they also rely heavily on our first party data, not the data in timescale.

We should refactor those to use more performant endpoints that query this data from timescale (example in: https://github.com/open-sauced/api/pull/593)