decouple data gathering from reporting

[ssbarnea]

There two very different activities: getting list of plugins and information about them and reporting.

Based on where we want to use the list of plugins we could generate rst, html, txt or json.

I am inclined to suggest using a simple JSON file as a database, mainly because we could store it inside the repository and still be able to humanly read it, or read diffs on incoming changes. Still, other approaches could also work, even a YAML file.

What do you think?

[mcsitter]

I think the integration to the main repository would be best and make the generation of the "database" unnecessary, especially since the rendered version in the docs is rather structured and could be scraped if necessary. It also would be redundant to the content that is in the rst file.