Fix #15: http/https inputs

[ejulio]

Fix #15
This PR adds two new schemes to spider-feeder, http and https.
They just open URLs using urlopen from urllib.request.
No extra data or certificates are allowed (maybe this is something that can be set somehow, but I didn't find a use case for it so far).
If there are query string parameters to be set, they can be formatted by the spider, as %(param)s are allowed in SPIDERFEEDER_INPUT_URI.

Also, I thought to include some transformations for common URLs.
For example, Google Drive displays the URLs like https://drive.google.com/file/d/<file id>/view?usp=sharing, but the actual URL for download is https://drive.google.com/uc?export=download&id=<file id>. To avoid requiring the users writing this piece of code, the transform could go inside the open method, what do you think?
At least, if not going there, and the user must write it explicitly in the spider, provide some utility function like get_google_drive_url(url) that would return the proper one and the user could use it when required.

[Gallaecio]

Regarding Google Drive, I feel like this could be covered in the documentation instead. But I guess a utility function for it is OK.

[ejulio]

@Gallaecio , I created an issue for the utility function #23