Importer should handle imports that are URLs

[camh-]

The importer handles file and https schemes, mapping normal file paths and netpaths onto these schemes. As the user of the importer can provide their own fetcher, they could handle other schemes themselves (such as internal: for wrappers around native functions for example).

The importer should check if files being imported have a scheme, and if so, assume they are absolute and not run them through the search path. It should also check for a scheme on the source parameter to Import to correctly get the directory of the source so scheme-relative imports can be done. It should also properly concatenate imports with elements of the search path that have a scheme (although I think that just works already as is).