[ENH] Improve reproducibility

[arnaudbore]

Type of improvement

So far here are the automatic modifications:

• if MRTRIX:
  export OMP_NUM_THREADS=${task.ext.single_thread ? 1 : task.cpus}
  export MRTRIX_RNG_SEED=${task.ext.mrtrix_rng_seed ? task.ext.mrtrix_rng_seed : "1234"}
• if SCILPY:
  export OMP_NUM_THREADS=${task.ext.single_thread ? 1 : task.cpus}
• if ANTS:
  export ITK_GLOBAL_DEFAULT_NUMBER_OF_THREADS=${task.ext.single_thread ? 1 : task.cpus}
  export ANTS_RANDOM_SEED=${task.ext.ants_random_seed ?: "1234"}

I still need to check ressources for eddy/topup/fodf_msmt/noddi/freewater.

If submitting a new module or fixing a bug, please use the appropriate template.

• Documentation
• Development tools (e.g. linter, formatter, etc.)
• Development container
• Global update (please specify)
• Other (please specify)

Describe your improvement

Write a clear and concise description of what the improvement is.

Describe how to test your improvement

Provide a full step-by-step guide to test your improvement.

Checklist before requesting a review

• Ensure the syntax is correct (EditorConfig and Prettier must pass)
• Run the test suites if your changes affect any module
• Regenerate the Poetry lock file if you have updated the dependencies
• Ensure the documentation is up-to-date

[gdevenyi]

Here's claude's rundown of common thread environment varibales:
https://claude.ai/share/041b8d97-a7d5-4544-88b7-dc761565a1f2

[gdevenyi]

Based on my reading here, I think:
OPENBLAS_NUM_THREADS=1 is needed anywhere scipy/numpy is used

TORCH_NUM_THREADS=1 is needed for tractoracle

This PR is really invasive, is it possible to set environment variables globally so we can more easily catch everything?

[gagnonanthony]

Great job! I noted the places where the seed arguments were not in the meta, otherwise minor comments here and there 😃

[AlexVCaron]

Based on my reading here, I think: OPENBLAS_NUM_THREADS=1 is needed anywhere scipy/numpy is used

TORCH_NUM_THREADS=1 is needed for tractoracle

This PR is really invasive, is it possible to set environment variables globally so we can more easily catch everything?

Applying those globally would require a nextflow.config file. There you can define env (top scope, not in process) taken into account by all modules at runtime. The problem right now is this file needs to be define at the highest level (mostly pipeline), because no importing config from modules or subworkflows is not part of the pipeline template, nor automated by nf-core commands.

[gdevenyi]

You're inconsistent here between ants_random_seed and ants_rng_seed, and even random_seed in registration/ants

[gagnonanthony]

LGTM! 💯