`mergeTranscript` duplicates sub-features if single transcript is passed to multiple times.

[hargup]

{
    "end": 46,
    "start": 1,
    "strand": 1,
    "subfeatures": [
        {
            "data": {
                "end": 46,
                "start": 1,
                "strand": 1,
                "type": "CDS"
            }
        },
        {
            "data": {
                "end": 46,
                "start": 1,
                "strand": 1,
                "type": "exon"
            }
        }
    ],
    "type": "transcript"
}

When a transcript created from the above JSON object is passed to EditTrack.mergeTranscript in form mergeTranscript(refSeq, [transcript, transcript]) we get a transcript corresponding to the following JSON object. Notice that we have two exons and two CDS which are identical.

{
    "end": 46,
    "start": 1,
    "strand": 1,
    "subfeatures": [
        {
            "data": {
                "end": 34,
                "start": 1,
                "strand": 1,
                "type": "CDS"
            }
        },
        {
            "data": {
                "end": 34,
                "start": 1,
                "strand": 1,
                "type": "CDS"
            }
        },
        {
            "data": {
                "end": 46,
                "start": 1,
                "strand": 1,
                "type": "exon"
            }
        },
        {
            "data": {
                "end": 46,
                "start": 1,
                "strand": 1,
                "type": "exon"
            }
        }
    ],
    "type": "transcript"
}

[yeban]

@hargup This one is critical. Please could you address this in a PR right away?

[yeban]

I had a look. The bug can manifest itself in several ways and not just as duplicated subfeatures if the same transcript passed (which will never be the case outside test suites). For instance, if two overlapping exons were merged, Afra will still consider overlap junction as a splice point when it shouldn't.

[yeban]

Thanks a bunch for reporting this issue. It was an important catch.