Preprocessing Performance

[jbusecke]

Opening this to follow up on #253. I have so far not really given that big of a priority to the performance of the preprocessing module, but I think it would be good to

1. Benchmark this, in order to detect performance degradation
2. Eventually spend some time to really optimize the time the preprocessing takes, since even small changes can sum up quite quickly when lazily reading in many hundreds/thousands of datasets.