Setup: Without increasing RAM limit the step for index generation will fail in Arvados

[zoe-translates]

In the Setup "episode", there are instructions for generating the STAR index from the input by running the STAR-Index.cwl workflow. The computation requires substantial RAM -- and without Resource Requirement, the container will run with bare minimal RAM quota and be killed in a short time when running on Arvados. (This particular task has indeed been used as an example to illustrate the point).

In the "Arvados" version of the page, should we include some ResourceRequirement for ramMin in a modified CWL file that we ask the learner to run? (In the "Generic" version the workflow is under the control of the user's local environment, and it should be fine as long as there's enough RAM available.)