The use of correct junction to exon differences in coordinates in createExonsByReadClass could introduce errors

[cying111]

When trying to run processReads on one of the samples, I found that this step in contructReadClasses might introduce errors for very small exons (1bp exon).

Before correcting: [3] 12 49129573-49129574 + After correcting: [3] 12 49129574-49129573 +

Currently this is causing errors when output passed to scoreReadClasses, where unlistIntrons would ignore this exon by giving one less intron, different from number of exons -1 and therefore causing issues to isReadClassCompatible.

I changed the code in isReadClassCompatible a bit for runs to pass through when this happens.

But I think maybe we should note this point down to see whether changes need to be made for such 1 bp exons.