Requested node configuration is not available error when running Dakota parallel evaluation tiling approach.

[CFDBloke]

I am attempting to run Dakota in serial mode using a sampling method to evaluate function responses. Each evaluation requires two codes to be run in parallel. Each code only needs one processor to run, but the codes must be on two separate processors for the coupling to work properly. Normally, to invoke a single run of these two coupled codes, a command would be issued taking the form:

srun --multi-prog -n 2 -K1 -W1 srun.conf

where srun.conf designates the command to be run on each processor (i.e. code_execution_path args). Obviously, I can't use this command in Dakota since it needs to be able to choose on what processors to run each evaluation. Therefore, I've been using a multi-instruction mpirun command in my analysis_driver file to invoke each evaluation, using this evaluation tiling example as a template. The run command in my driver.py file takes the form:

cmd_args = ["--bind-to", "none",
                     "*code1_execution_path code1_args* : *code2_execution_path code2_args*"]
command = (2, cmd_args)

parallel.tile_run_dynamic(commands=[command]

I've tried running both dynamic and static scheduling configuration. The interface section of my Dakota input file looks like this:

interface
    fork
       asynchronous
       evaluation_concurrency = 18 # Each node has 36 processors, allowing 18 simultaneous evaluations on 1 node.
       # local_evaluation_scheduling static # Uncommented when running static scheduling
       analysis_driver = 'dynamic_driver.py'
       parameters_file = params.in
       results_file = results.out
       file_tag
       file_save

     work_directory named 'workdir'
     directory_tag
     directory_save
     
     copy_file = 'templatedir/*'

The job submission script to our HPC looks like so:

!/bin/bash
--#SBATCH --partition=parallel
--#SBATCH --nodes=1
--#SBATCH --ntasks-per-node=36
--#SBATCH --time=48:00:00
--#SBATCH -o outfile
--#SBATCH -e errfile

module purge
module load *all_modules_for_evaluation* python dakota/6.19

dakota -i dakota.in

rm -f ${HOME}/.DakotaEvalTiling/${SLURM_JOBID}

When I submit the job script to the hpc, the Dakota pre-processing of the input decks for each evaluation code works fine, but then I get the following outputted to errfile multiplied over 18 lines:
srun: error: Unable to create step for job xxxxxx: Requested node configuration is not available

I think the problem lies in the invocation command I'm applying within my analysis driver file. So my question is whether you can see if there is something obviously wrong with my inputs here?