Fix number of cpus for modules with piped tools

[JoseEspinosa]

Split the number of CPUs in modules with two tools in a piped command and each of them with an explicit declaration of the CPUs to be used. See #498

PR checklist

Closes #498

• This comment contains a description of changes (with reason).
• Ensure that the test works with either Docker / Singularity. Conda CI tests can be quite flaky:
  • PROFILE=docker pytest --tag <MODULE> --symlink --keep-workflow-wd
  • PROFILE=singularity pytest --tag <MODULE> --symlink --keep-workflow-wd
  • PROFILE=conda pytest --tag <MODULE> --symlink --keep-workflow-wd

[ewels]

LGTM 👌 Just need to fix the tests..

[maxulysse]

<3

[JoseEspinosa]

LGTM 👌 Just need to fix the tests..

These are the infamous conda tests... I can not also resolve the environment locally, so I guess that as usual, since docker and singularity are passing we can ignore it... 😓

[ramprasadn]

Sorry for resurrecting this thread four months after it was closed, but I wonder if splitting by 2 affects the performance of the aligner a bit.

Looks like when we have a piped command, its almost always an aligner paired with samtools view, and samtools view is not a resource intensive process. In that case, the aligner should be getting more resources instead of samtools, right? What do you guys think?