`Extractor`: implement `usage` and `installation` schemas

[PeterKraus]

This PR re-implements the usage and installation attributes of an Extractor, so that the execution via the api can be fully self-contained using information from within the registry. Closes #19 .

• [x] usage reimplemented:
• [x] usage:method child entry uses an enum (currently cli and python are the only options)
• [x] usage:setup child entry for defining e.g. import extractor or import package.extractor statements
• [x] usage:command child entry has a defined set of template words (e.g. $filetype) and validates its syntax
• [x] installation implemented:
• [x] installation:method child entry uses an enum (currently pip and conda are the only options)
• [x] installation:requirements child str entry optionally contains contents of an "requirements.txt" or yaml file
• [x] installation:packages child list[str] entry containing packages (versioned, git+https, etc.) to be installed by installation:method

[PeterKraus]

@ml-evs this should be in a better shape now, please have a look once you have time. One thing that's still open is the set / format of "keywords", related to the Extractor:usage:command field. We can either push this PR as is and leave #19 open, or resolve #19 there and then add a longer docstring for that attribute here.