Whether to enable a workaround for document classifier timeouts.
Paperless uses OpenBLAS via scikit-learn for document classification.
The default is to use threading for OpenMP but this would cause the document classifier to spin on one core seemingly indefinitely if there are large amounts of classes per classification; causing it to effectively never complete due to running into timeouts.
This sets OMP_NUM_THREADS to 1 in order to mitigate the issue. See https://github.com/NixOS/nixpkgs/issues/240591 for more information .