HTTP errors and download picker

The upload picker was sadly a bust, because Firefox does not give you the full path, just the filename.

The download picker is also a bit tricky. It doesn't return a path either, so you can't ask it for a path and give the path to siad. Instead, siad automatically downloads files to webpages/downloads/. Then, the download picker is used to "download' (actually copy) the file to its intended destination. This creates an efficiency concern, because you wind up with two copies of the file. A move would be more efficient, but it's not possible using this scheme. Slightly better would be to delete the file after copying. This would be implemented via a new /delete?filename= API call.