Fix large downloads (#105)

[t8y8]

I've got a working solution for #105

Requests provides a way to just get the response headers back and then iterating over the content in chunks. This lets us stream to disk instead of trying to hold large workbooks in memory.

(See http://docs.python-requests.org/en/master/api/#requests.Response.iter_content and http://docs.python-requests.org/en/master/user/advanced/#body-content-workflow)

As a proof of concept I extended our wrapper methods to allow per-call http options (because we don't need to stream every request) and then tested it on workbook download with the script below.

If this approach works, I can clean it up and add it to data source download as well.

But a few questions:

1. Are we ok with the way I extended our request-get-wrapper function? An alternative is to have a special get_download_request instead
2. I need to make sure the response always gets closed once we use stream=True I think I'm right but may want to just wrap it all in a context manager or try/except
3. Preferences on a chunk size? (I did 4MB because it's usually safe)

import tableauserverclient as TSC

authz = TSC.TableauAuth('user', 'pass')
server = TSC.Server("http://server")

server.auth.sign_in(authz)
print("Signed in: " + server.auth_token)

#workbooks = TSC.Pager(server.workbooks)

big_wb = "FIND ONE OR USE CODE TO FIND ONE"

print("Downloading workbook: " + big_wb)

server.workbooks.download(big_wb)

print("DONE")

[t8y8]

Added datasources and wrapped it all in closing() so we can be sure the connection gets closed.

I went ahead and tested on large and small items on an internal server, and realized that the chunk size didn't actually seem to affect the download time that much -- so I went with 1KB, what most examples on the internet do.

I verified python process memory doesn't balloon while downloading a 3GB file, and that it did without the fix.

Side note: Currently download will just overwrite the filename -- not sure if that's what we want as the default behavior or not

[LGraber]

This is really clean. I like it very much. 🚀

[LGraber]

Very nice