[srivarra]

Related issue

Needed for Biohub/biahub#462, which runs Richardson-Lucy deconvolution in biahub's mantis-v2 reconstruct step.

Summary

apply-inv-tf and reconstruct ignore the config's device. apply_inverse_transfer_function_single_position calls the phase and fluorescence models without it, so they run on the CPU whatever the config says. A Richardson-Lucy config with device: cuda allocated no GPU memory on an A40.

This passes settings.device to the phase and fluorescence models. The birefringence models take no device and still run on the CPU.

Testing

• New gpu test test_apply_inv_tf_runs_on_the_config_device, for phase and for fluorescence with RL. With device: cuda the reconstruction allocates GPU memory and matches the CPU result.

Checklist

• I added or updated tests, or this change does not require tests.
• I updated the documentation, or this change does not require documentation.
• I confirmed this change does not include confidential or sensitive information.