Option to specify Madrigal file directly

[ljlamarche]

Currenly, when a Madrigal dataset is requested, MadrigalWeb is used to first query the server and identify the file name, then download the file. This work, but it can be slow. If the user has the file downloaded locally, they should be able to just provide the file path, which should be much more efficient