Refactor 'list' and 'prune' operations on remote repositories to run on the server side

[Ernest0x]

The list and prune operations on relatively large (in size and number of archives) remote repositories are taking too much time and create a lot of unnecessary traffic, which would be avoided if they were implemented as server-side operations that return only a small-sized response to the client. In addition, the slowness of the current implementation may be revealing some other problems as described in #161.

[ThomasWaldmann]

If you have an encrypted repo, attic on the server does not have the key, thus it can't see the informations required for list and prune.

But see #167 for another idea how to speed up list and prune.

Close?