Infer storage_type from file name (always?)

[valeriupredoi]

In #89 I had to introduce a new kwarg to Active -> storage_type to have correct open/retrieve operations depending on the type of storage (POSIX vs S3 now) - storage_type should/could be picked up from file name as @davidhassell suggested offline - but we can change that later (am not 100% sure we should, though, since future possible storage types may not necessarily be guessed from file name) - this issue is to raise our awareness as we go along.

[markgoddard]

You might want to reconsider how the S3 access is configured - is global config in config.py the most appropriate approach? I used it as a fast way to get things working.

Would it be better to pass in the URLs & credentials etc. to the Active object's constructor? Or possibly abstract them into a Storage object that provides a reduce_chunk interface and "knows" how to access the object in storage.

[valeriupredoi]

that's a very good point and question! I am thinking we should probably discuss what other types of storage we want to support, and then see what's the best way to have them configured at runtime - I was too thinking about config.py when you opened the PR a few weeks ago and was leaning towards a yaml config file (still sort of an externally loaded configuration) but a config object might be a better approach. For now it does the trick as interim solution 👍