Fix identification of full length informative reads when using single end reads

[timothymillar]

If single end reads are used then TEF should still be able to use soft-clips (but not full danglers).

At the moment this fails though because the extraction process incorrectly identifies unmapped single end reads as having a mapped pair.

This happens because read.mate_is_unmapped (used here) always returns False for a single end read but it is currently used assuming a paired end read in which case it only returns False if the mate is unmapped.

Because of this the extraction script recognises an unmapped single read as having a mapped mate an therefore treats this read as an informative full length dangler.

Later on the script calls read.next_reference_name to get the name of the element that the mate read mapped to but this returns None for a single end read which ultimately results in a pysam error:

Traceback (most recent call last):
  File "/workspace/appscratch/miniconda/cfltxm_tefingerprint_v0.3.2/bin/tef-extract-informative", line 12, in <module>
    main()
  File "/workspace/appscratch/miniconda/cfltxm_tefingerprint_v0.3.2/bin/tef-extract-informative", line 8, in main
    Program.from_cli(sys.argv[1:]).run()
  File "/workspace/appscratch/miniconda/cfltxm_tefingerprint_v0.3.2/lib/python3.7/site-packages/tefingerprint/_applications/extract_informative.py", line 249, in run
    self._run_pipeline()
  File "/workspace/appscratch/miniconda/cfltxm_tefingerprint_v0.3.2/lib/python3.7/site-packages/tefingerprint/_applications/extract_informative.py", line 197, in _run_pipeline
    self.mate_element_tag)
  File "/workspace/appscratch/miniconda/cfltxm_tefingerprint_v0.3.2/lib/python3.7/site-packages/tefingerprint/_applications/extract_informative.py", line 512, in tag_danglers
    read.tags += [(tag, mate_element_dict[read.qname])]
  File "pysam/libcalignedsegment.pyx", line 2763, in pysam.libcalignedsegment.AlignedSegment.tags.__set__
  File "pysam/libcalignedsegment.pyx", line 2571, in pysam.libcalignedsegment.AlignedSegment.set_tags
  File "pysam/libcalignedsegment.pyx", line 418, in pysam.libcalignedsegment.pack_tags
ValueError: could not deduce typecode for value None

This can be avoided by checking if a read has a pair with read.is_paired before trying to identify if it is a full-length informative dangler.
Cheching for soft-clips should always use read.reference_name instead of read.next_reference_name to avoid this issue

This issue is already partially resolved in the dev version which prioritises soft-clips over full length danglers because the use of full-length reads can be disabled when using single end reads.