Reconstructed images should be linked with transparent description on how they were generated.

[MauroLeidi]

I propose we add functionality to save reconstruction metadata in addition to the reconstructed image files. This metadata should capture all key steps and parameters used during the reconstruction process, enabling reproducibility and easier debugging.

The metadata could include, for example:

• The raw data file name (optionally with its hash) and the AcquisitionParam used in the rawdatareader.
• The trajectory_type.
• The volume element argType.
• Parameters used for coil sensitivity estimation, including the file names (and optional hashes) for the body coil and head coil, as well as the AcquisitionParam for their rawdatareaders.
• The reconstruction function and its parameters.
• The normalisation value used for the raw data (optionally including the ROI used to estimate it).
• The binary mask, along with any manually applied changes—specifically, the two threshold values and the bounding box dimensions.
• I envision this metadata being saved as a .json file alongside the reconstructed image.

(This could also lay the groundwork for supporting config files to control the reconstruction process. Researchers often need to run many similar reconstructions, and re-entering the same parameters repeatedly (e.g., in the rawdata readers or for ROI estimation) can be tedious. Being able to estimate these once and reuse them in subsequent runs would significantly streamline the workflow. That said, this would be a more complex feature with additional tradeoffs to consider.)