At some point we thought it would be a good idea to be able to run tests like 

    pybot http://my.server/tests.html

Even though we haven't "advertised" it, we still have this support. The
biggest reason why it isn't really useful are:

- You can only execute single test case files, not directories. We could,
of course, add support for following links from index.html files but that
wouldn't be trivial.
- You need to download needed libraries anyway. We could, of course, find
libraries that are imported using physical paths, but that would not be
enough as a general case.
- It is easy to create a separate tool for downloading test case files and
everything else needed. This kind of orchestration capabilities are better
kept out of the core framework. 

Supporting test cases files, resource files, and variable files that use
URLs makes many parts of the code really complicated, because we always
need to take into account that the needed file might need to be downloaded
first. Removing this not-really-useful functionality makes the code easier
to maintain and makes further refactoring easier.