[FEATURE] make picard markduplicates smarter

[bwlang]

Is your feature request related to a problem? Please describe

picard markduplicates does not follow current best practices with regard to duplicate type tagging, UMI usage, and "free" index creation

Describe the solution you'd like

I'll add some logic to the module to detect presence of RX tags, auto-create an index (which you get for free with --WRITE_INDEX), and illumina instrument type detection.

Describe alternatives you've considered

I was worried that
according to https://nf-co.re/developers/modules#inputoutput-options
such smarts should not be added but after a bit of conversation in slack, it seems a PR might be welcome.

some "pros" to having this logic:

• encourages scientific best practices
• bams 'in the wild" will be more accurate
• very low risk of doing something wrong

some "cons"

• more responsibility to keep up with best practice (not sure this is really a con... seems like a responsibility for a workflow maintainer anyway)

Are there other cons that I'm not considering?

Additional context

No response

[famosab]

@bwlang Do you still plan to open a PR regarding this issue? :)

[vagkaratzas]

I don't see anything related to --WRITE_INDEX in the module, so I guess this has not been implemented yet. @bwlang are you planning to do that or can we close this issue? Else maybe provide an example on the usage so somebody else can be hinted on how to implement this?

[bwlang]

@vagkaratzas : looks like the PR is passing tests and is ready for review.

[vagkaratzas]

Hmm since this PR (i) has added manual code in the script block, (ii) the md5sum of the existing outputs seems to have changed, and (iii) the module is already used in 14 nf-core pipelines, I would ask the opinion of the actual pipeline maintainers that use this. Tagging a couple here:
@Joon-Klaps @JoseEspinosa @phue @zachary-foster