[ENH] Need for more comprehensive integration testing

[fabianegli]

@ypriverol Do you have a set of SDRF files for experiments of all different kinds of expected flavors?

If not we should systematically generate SDRF example files with all the different labelling techniques, fractionations, biological and technical replicates, file names (spaces/weird characters, if allowable) to have test cases for at least the most common experimental setups. Based on those we can then generate examples of SDRF files that do not comply with the standard, but should be readable nonetheless and some that are wrong in various ways we expect users to mess up writing SDRF files with common tools or even by hand.