Add functionality to rtweet to sample a portion of the twitter graph

[alexpghayes]

Most people build graphs based on word usage because there are strict API limits on querying friends of a given user. Nonetheless, sometimes you want to do this. We should automate this process and submit it as a function rtweet. Since it has to be long running for any particular use case, it's worth thinking about fault tolerance. Also, since some people have a million friends / followers it's easy to end up searching the whole twitter graph if you don't limit the depth of the search / number of returned friends in some way. Some configurable settings on how to do this would be nice.

Finally, some functions for visualizing the resulting graph / converting it into a sparse adjacency matrix / igraph / tidygraph object could be nice.