Reliable ways to quantify package usage (specifically core astropy)

From Astropy Coordination Meeting 2021. When the discussions happened, it was about `astropy` core library, though the solution might be applicable to any number of Python packages in this Project. From the meeting notes:

* Still difficult to measure “usage” of `astropy` (citations is biased, visits to astropy.org vs docs is biased, etc.)
* Might be useful to publish some plots from the web analytics for the project to use. This is something people can understand pretty well, esp. Trending.
* @adrn: Agreed! I wanted to make a dashboard with all of this info, updated automatically, eventually at `astropy/astropy-sustainability` ([initial attempt here](https://github.com/adrn/astropy-sustainability))
* @bsipocz: Number of issues or PRs per subpackage is a coarse indicator as well
* @adrn: I wonder if it would be useful for usage to compare the number of issues raised by people who have submitted PRs to astropy vs. not (per subpackage)
* @pllim: Like this?


Currently, everyone seems to have their own favorite scripts to run for usage stats; e.g.,:

* https://github.com/astropy/astropy-procedures/tree/main/visualizations_demographics (Had not been updated for 3 years, last updated by @keflavich.)
* https://github.com/adrn/astropy-sustainability (As mentioned by @adrn above.)
* https://github.com/bsipocz/astropy-graphs/tree/extend_plots (Original scripts from @astrofrog modified by @bsipocz.)
* https://github.com/pllim/astropy-graphs/tree/lim-edit-2021 (Yet another modified version by @pllim.)
* Scripts by @taldcroft (private communications).

Start by reviewing the scripts in astropy-procedures/visualizations_demographics, astropy-sustainability, and the astropy-graphs branches linked in the issue, then compare the usage signals and plots they already provide. Done would require an agreed reliable usage measure and a reproducible, automatically updated dashboard or set of plots for core astropy.

Written by the indexing model from the issue text.