tiny-count: support a sequence-based counting mode when GFF files aren't provided

[AlexTate]

The pipeline currently requires the user to specify at least one reference GFF. However, there may be instances where they don't have or want to use genomic/reference annotations (.fa/.gff). For example, instead of a genomic fasta file they may instead want to use a synthetic fasta with known sequences to count against.

The behavior of tiny-count will be changed to allow for this. If a user does not specify any GFF files:

• Each entry in the user's reference genome file is treated as a "pseudofeature", where the description line is treated as the pseudofeature's ID. The implementation will infer the ID and length of these pseudofeatures from @SQ flags in input SAM files.
• A GenomicArrayOfSets will be populated with the pseudofeature coordinates just as they are with features
• Accordingly, Stage 1 selection will be skipped in this mode

[AlexTate]

Reopening this issue to add support for the Classify as... values in this mode. They're illustrated as being part of Stage 1 selection so I didn't include them in order to avoid confusion and be consistent with "Stage 1 selection is skipped." With the current explanation of "these reference sequences behave like features that had matched every rule in Stage 1," I think we could include them without adding too much confusion.