Introduce CWL option for guessing secondaryFiles.

The gdc_dnaseq_transform GA4GH workflow has tools that require indexes be present in secondaryFiles for at least one of its inputs but does not define these in the input document. I think this workflow should not work - but cwltool allows this behavior so we should add a flag or something to planemo for attempting to detect this I think (we could be formal and look at the workflow or we could be fuzzy and just use the files in the directory alongside the input).

From Gitter:

>John Chilton Sep 14 15:59
Should I have to automatically infer secondary files as part of the cwl-runner interface based on the tool/workflow being targeted or are they mandated to be part of the input document?
> Michael R. Crusoe 04:52
@jmchilton Good question, I don't know off hand
> Brad Chapman 05:32
John, some runners infer them but you're meant to explicitly specify them in the CWL. I had some missing in bcbio generated CWL and when testing on DNAnexus (which did not infer) picked these up.
One cavaet is that right now there is no way to specify secondary files for record files, so you have to infer there.

Start with Planemo's cwl-runner interface and the gdc_dnaseq_transform workflow described in the issue. Clarify whether detection should inspect the workflow formally or infer neighboring files, then define how the new option should report or apply inferred secondaryFiles and what behavior confirms it is complete.

Written by the indexing model from the issue text.