I emailed MovieLens and got the following response. I think it may be better to find or create other .tsv/.psv files so we don't have to configure any accepting of licenses.
The data sets come with READMEs that outline the acceptable usages. Since you are with Apache, I believe you would not be using this for commercial purposes so you satisfy that clause. The main roadblock I see is what is involved with using the data set as a test resource. Would this mean the data would be hosted in an SCM somewhere for anyone to download?
Our LensKit framework uses the 100k data set as a test resource, but has Maven download it automatically after using a property to "accept" the license. If you were able to configure your testing environment similarly, then I think using our data would be acceptable.
University of Minnesota