If memory is specified, tasks will not run on the UGE engine

Bug report

Parameters in config h_rt/h_rss/mem_free should not be specified by Nextflow.
This is because some supercomputers are unable to accept these parameters. I am not sure if this is common or not.
However, at least on NIG supercomputer system( https://sc.ddbj.nig.ac.jp/ ), ques never run.
We need to use h_vmem instead of those parameters.

https://nfcore.slack.com/archives/CGRTMASKY/p1629489188025100

I solved the problem by this conversation's methods.

In config file
memory = null

Is there any way to replace this, for example by creating a new option?
Null sometimes causes errors in downstream workflows.

Expected behavior and actual behavior

Ques run immediately.

Ques never run.

Steps to reproduce the problem

Not works.

• memory = '16G'
• memory was not specified
  Works
• memory = null

Environment

• Nextflow version: 21.04.3
• Java version: 11
• Operating system: CentOS7.5

Start by tracing the UGE engine's handling of the h_rt, h_rss, mem_free, h_vmem, and memory configuration mentioned in the report. Reproduce the difference between memory='16G', an omitted memory setting, and memory=null on the stated environment, then verify that a supported configuration lets queued tasks run without introducing downstream null errors.