Add exclude subject/session option

[jdkent]

So that we don't redownload repeat sessions. The alternative solution is to create a session dictionary (like we did with the subject dictionary) to change cases where the extra session should be downloaded with the main session. The drawback is when there are overlaps in filenames (e.g. We collected half a functional run at the actual session and want to overwrite that scan with another that we collected at an extra session to make up for incomplete data).

So I think I'm going to go with the exclude option(s) for subject and session, as that will cover a variety of use cases.