input an image as FITS hdu object

[caseyjlaw]

I am interested in using pybdsf for source detection within a larger framework for image processing. I would like to run image generation/manipulation, then run source detection. Currently, pybdsf requires specifying an image file as input. This requires me to write a file after image generation and then read it back in for source detection, which is slower and less elegant.
From what I see, we could pass a FITS hdu as input. This could be implemented at https://github.com/lofar-astron/PyBDSF/blob/master/bdsf/readimage.py#L68 or at https://github.com/lofar-astron/PyBDSF/blob/master/bdsf/functions.py (in the read_image_from_file function).
Is there interest in a pull request to update the readimage module to support this use case?

[darafferty]

Sorry for the slow reply. Yes, I think such functionality would be very nice to have, and we'd be more than happy to accept a PR for it.

[caseyjlaw]

Thanks. I got this working in my fork by using copilot. However, I'm not very happy with it. The PR from my feature to main is at https://github.com/ovro-lwa/PyBDSF/pull/1/changes.
As you can see, it adds a lot of code to the __init__.py and manually sets the properties of an Image object. If you could suggest a better place to implement some of this, I could try again. I'd like to be sure the PR respects the logical organization that you have set up in the main branch.