refactor caching

[jstriebel]

• clear cache only for relevant datasets on /data/tiggers/clearCache/<organization_name>/<dataset_name> endpoint
• configure cache-size based on dataset
  The cache size should depend on:
  • number of datasets
  • per dataset
    • datatype
    • chunk_size
  • Probably it should be one global cache that is aware of the stored sizes.
• Consider Redis as a cache to enable multiprocessing.
  see https://redis.io/topics/config#configuring-redis-as-a-cache
• Caches might need to be invalidated if dataset config changes