Implement blob writing for azure storage

[rklaehn]

This required a signature change in the BlobWriter trait to return an Url instead of just ().

I also added a way to provide credentials via an environment variable AZURE_STORAGE_ACCESS_KEY.

Questions:

• should the blob bucket or whatever it is called be created if it does not exist? Currently the code assumes that it exists.
• should the URLs that the azure provider uses be changed to be relative? Currently the provider uses absolute URLs, so if you were to point the provider to a different account you would either have to modify the db, or things would stop working.

Current URL scheme:

https://eueast.blob.core.windows.net/blobs/bafkr4ifzfd6kvgwfd35v2iyz3gqjrtyjssgiqrgwxu25tegq352hwzzbju
        |----| configured in the provider
                                     |---| configured in the provider
              |--------------------| also implicitly configured in the provider by using MS

My proposal would be to use the url field in the db only in the case where the data is not in a canonical location that the provider can figure out using the cid, and even then only add the extra info needed. So in this case url would be just the file name to allow the case where we have an azure bucket with named files that we want to content address.

[cameronfyfe]

Answering questions:

should the blob bucket or whatever it is called be created if it does not exist? Currently the code assumes that it exists.
We can just assume it's been set up and exists for now

should the URLs that the azure provider uses be changed to be relative? Currently the provider uses absolute URLs, so if you were to point the provider to a different account you would either have to modify the db, or things would stop working.
This sounds good to me, makes more sense to construct full urls on the fly

[rklaehn]

Added some integration tests that just push some data to an azure bucket.

They are currently ignored since you need azure credentials.