Add tests that use real sequencing data as input

[matthuska]

Current automated tests are only using stubs. It would be very nice to use (small/tiny) real sequencing data for our tests to have higher confidence that the pipeline is doing what we want, and that we're not breaking anything when we make changes in the future. We need at least one Illumina and one Nanopore data set.

The data sets should be kept extremely small to make the tests faster and to make it easier to prepare expected output by hand.