Determine memory requirements for download

[johnbradley]

The current download logic may be importing the entire file into memory here:

dataverse-access/dv.py

Lines 73 to 74 in 7a67502

	response = data_api.get_datafile(file_id)
	f.write(response.content)

Determine if there is a streaming approach to reduce requirements.

[johnbradley]

The data_api.get_datafile(file_id) call will download the entire file and keep it in memory. There is an issue requesting the ability to add the streaming flag to avoid this un-necessary memory allocation: gdcc/pyDataverse#49

[johnbradley]

I am looking into using libcurl to handle streaming uploading and downloading via pycurl.
I had trouble installing pycurl with pip in multiple environments. It would install fine, but fail at runtime.
So using pycurl would likely require conda.