Sbatch fatal error when mem-per-cpu is included and not when not included

I am trying to run the demo mode of this pipeline from Nanopore using the apptainer profile on an HPC that uses the slurm scheduler.
If specify the --mem-per-cpu in the clusterOptions, I get the sbatch error:
--mem --mem-per-cpu, and --mem-per-gpu are mutually exclusive.
This makes sense because the .command.run script that NF generates for the slurm executor includes:
#SBATCH --mem 2048M

However, if I leave out the --mem-per-cpu in my clusterOptions, then I get the fatal sbatch error that --mem-per-cpu is missing:

demo_HV_NF_19774085.out.txt (23.9 KB)
nextflow.log (46.6 KB)
nextflow.config (3.4 KB)

I am including my nextflow.config file. and log file.
I am restricted to using Nextflow version 22.6.10 , so I can’t use the executor.perCpuMemAllocation option.

I have heard from IT at my HPC. The cluster requires the --mem-per-cpu flag, but Nextflow will always put the --mem in the command, so I cannot run any commands. How can I change the behaviour of the executor so it does not include the unnecessary flag?