Allow choice between 3' and 5' read ends

[bmmalone]

Currently, the code is hard-coded to use the 5' ends of riboseq reads. However, for some sequencing protocols (mostly for prokaryotes), it has been shown that using the 3' end is better. The logic for end selection is entirely in the riboseq-utils package. In particular, it appears to only be used in:

• extract_metagene_profiles.py: around line 114, the call to bam_utils.get_five_prime_ends
• ribo_utils.get_p_sites: around line 510, around the message "Discarding 3' end of reads"

It appears the choice could be added by passing a relevant flag to these locations in the code.