Bugfix 2

[mkabza]

Fix for the issue #2, concerning @lingminhao and @mcortes-lopez

Implemented changes:

• parsing BAM files has been made more stringent by disallowing secondary alignments, as well as reads that are duplicated or don't pass QC criteria
• inter-chromosomal alignments are now automatically removed from the analysis
• added the max_intron_length (default 5e6) argument to the prepare_transcripts function, which should lower FDR in some cases
• added new assertion to the prepare_transcripts function, which should catch problems with non-unique intron_positions values in input data
• updated the unit tests to reflect the introduced changes and re-built the vignettes & the manual