Improved cache handling

[pitkant]

• Use md5 hash as cache file name to check if a file has already been cached
• Calculate md5 hashes pre-emptively for filters = NULL queries to check if a superset of desired dataset already exists in cache
• Allow caching for all datasets, including filtered datasets, instead of just bulk downloads

Fixes issues in #144, #257 and #258.