Executor run CLI not respecting --qc-compute-n-cores

[jthorton]

Running the example in the docs using bespokefit version 0.1.0 installed via conda-forge sets up 2 qc workers but gives them both all available cores rather than the 8 specified via the CLI.

[SimonBoothroyd]

@jthorton I'm not sure I can reproduce this. When I run:

openff-bespoke executor run --smiles "CC(=O)NC1=CC=C(C=C1)O" --workflow debug --directory bespoke-fit --n-qc-compute-workers 2 --qc-compute-n-cores 4

for example i see

[2022-03-01 09:49:31,489: INFO/ForkPoolWorker-1] openff.bespokefit.executor.services.qcgenerator.worker.compute_torsion_drive[d8badf49-9ee2-4129-ac39-bf0cd51db32c]: running 1D scan with {'ncores': 4, 'nnodes': 1, 'memory': 32.0}
[2022-03-01 09:49:31,496: INFO/ForkPoolWorker-2] openff.bespokefit.executor.services.qcgenerator.worker.compute_torsion_drive[cbdfa4aa-d66f-4d32-b0c6-11299046ff71]: running 1D scan with {'ncores': 4, 'nnodes': 1, 'memory': 32.0}

in the celery-qcgenerator.log file.

[jthorton]

@SimonBoothroyd thanks for testing this the issue only seems to affect Linux and works as expected for me on mac.