comprna/reorientexpress: first release

This is the first release of ReorientExpress, a program to create, test and apply models to predict the 5'-to-3' orientation of long-reads from cDNA sequencing with Nanopore or PacBio using deep neural networks for samples without a genome or a transcriptome reference. ReorientExpress implements two Deep Neural Network models: a Multi-Layer Perceptron (MLP) and a Convolutional Neural Network (CNN), and it uses as training input a transcriptome annotation from any species or any other fasta/fasq file of RNA/cDNA sequences for which the orientation is known. Training or testing data can thus be experimental data, annotation data or also mapped reads (providing the corresponding PAF file).