Support more than 5TiB max blob size

[xinaxu]

Zenko service has hardcoded constants to be as close as AWS S3 as possible. Those constants could be overriden to support more than 5TiB blob size.

We also need to make sure any client side tool, i.e. AWS S3 CLI or AWS S3 SDK supports larger part number and part size.

[elijaharita]

Modifying the constant should be easy enough, but how would we go about making sure this works?

I'd need access to a machine with like 16TiB storage to actually test on a devnet. Otherwise I'd just be guessing whether things are working or not.

Unless I can test on an existing live instance and just stream 5TiB of data from urandom ??

[rvagg]

How about we just mock out the backend, that would then set us up for testing other types of complex behaviour. Either we make an /integration/mock so we can have an instance with an alternative backend, or we make a mock SingularityAPI which we can plug in to pretend that there's a real singularity there. A mock singularity might end up being very useful, although a little more work. But for the purpose of testing this, the amount of the API we'd need to implement is fairly minimal.