As needed per #151, I added parsing of atom tags in RSS.

This adds a mechanism to add a parser for certain extension namespaces. This is now only implemented for calling Atom from RSS. This avoids re-handling Atom in RSS by calling the Atom parser directly.

The implementation tries to be very agnostic: the RSS parser knows nothing about Atom. I played with the idea of removing the abstraction and couple them tighter, but the result was not that different -- so I settled with the nicer agnostic one.

The translation phase, of course, is not agnostic anymore, because it has to know what to put where.

Remarks:

• This implementation currently only supports calling the external parser on the item level. Adding parsing on channel level is trivial, the translation phase then is more involved, because currently everything expects an *atom.Entry.
• The initialisation of the inner atom parser might not be completely correct (see atom/parser.go#ParseAsExtension), because I do not know the idea behind base and the url stack.
• Due to the agnostic ways of the approach, the JSON decoder does not precisely know what to do with the extensions. Thus, for the tests, the extensions of the actual object take a JSON-detour (i.e., en- then decoded) to match the expected.
• The translation of atom-tags is currently only added for author and updated. If more are needed, they need to be added.

Resolves #151