Redesign the web service harvester and create design documentation
A document on how to do the harvesting to fit inn with the new dynamic actor
loading, annotation, etc. The harvested services should not be kept in the
repository, and they should be updated everytime the Kepler client is loaded. We
probably need to have a metarepository in the end to keep track of the
repositories and keywords that were searched and also collect some information
on the user's interest.