* greps for manpages with nix-locate
* fetches the nar archives non-recursively
* extracts manpages
* converts to html

This involved implementing a multiprocessing limit and lru file cache using bash...