Fix Manta output

[maxulysse]

Some output was not there anymore due to #579

[maxulysse]

We had $filename as an output, and it should have been filename.
I also took the opportunity to reduce PublishDir directive.

[maxulysse]

Do not merge, yet, there is still some issues to figure out...

[maxulysse]

OK, so in fact as long as a file is an output, it will be in the work directories.

So I present you this final fix to our problem.

The idea was to use the join operator to phase the bams/bais and the RunManta output (small) without feeding these into RunManta.

In order to do that I choose to reorder the items in our input/output.
Instead of: idPatient, idSampleNormal, bamNormal, baiNormal, idSampleTumor, bamTumor, baiTumor
we now have: idPatient, idSampleNormal, idSampleTumor, bamNormal, baiNormal, bamTumor, baiTumor

I think we have some other process that have an output that will be in the work directory when using scratch, so I'll need to fix that as well.

I do think that we're using the now depreciated phase operator, so it'll be fixed as well.

[maxulysse]

Finally remapping the channels before joining so we don't touch other parts of the script.

So finally what we're now doing is not using bam as an output in Manta, but joining a channel containing the bam to the StrelkaBP output.

I tested it, and it seems to be working ;-)