Create gene and transcript length tables

In order to run the PCA analysis (#140 (closed) !77 (merged)) we need to have two tables with the gene and transcript length.

Initially, I thought to randomly select a sample and take the corresponding effective length. The problem is that for paired-end data the effective read length is always different. This should also vary between paired-end and single-end data as for the latter we specify manually the fragment length.

An alternative option is to use the normal length of the genes and transcripts. Do you think that it's ok or do you have an alternative suggestion? @zavolan