File URI regex does not work with GNU grep, blows up workspace

[suskin]

Hey guys!

The download script completely fubars the destination directory on machines using GNU grep, especially for accounts with lots of files to download. The script creates a lot of files whose filenames are long JSON strings.

The reason is the FILE_URI_REGEX which is passed into grep does not work with GNU grep; for GNU grep, the ungreedy qualifier ? requires grep -P rather than grep -E to function. See this stack overflow question for some discussion.