Add new -J flag for enhanced DICOM metadata extraction

[hanayik]

Add new -J flag for enhanced DICOM metadata extraction

Summary

This PR introduces a new -J command-line option to dcm2niix that enables more DICOM metadata extraction to JSON format, going beyond the standard BIDS sidecar fields.

[neurolabusc]

@hanayik this is an interesting PR. It has an elegantly minimal implementation and can extract a lot of code. However, I do not think it should be included in the main distribution, or if it is it should only be bundled with a compiler directive. A few comments:

• It makes pretty strong assumptions about the DICOM data being valid, and will falter with several violations of the standard that are common in the wild.
• The behavior for implicit VR data is particularly poor: without a DICOM dictionary or any vendor specific details it generates invalid JSON files (e.g. repeated keys).
• The style of JSON is inconsistent with the other JSON export (e.g. double spaces instead of tabs for white space).
• In general, JSON is a a poor choice for large amounts of meta data. While it is human readable the parsing is slow and the use of ASCII for floating point numbers is both slow and lossy. Given that a lot of BIDS datasets are part of huge analyses, I would keep the main JSON lean with minimal extra details.
• Now that the BIDS steering committee has an established method for improving the specification, I would prefer if new JSON tags are first ratified by that organization before being introduced into only a single software implementation. I do not think I should be gatekeeper for determining what key:value pairs are required.
• In particular, your tool seems to generate repeated keys with varying values for nested DICOM tags, common in enhanced DICOM. I think there needs to be some way to preserve the structure of this information, not merely the names.

I do think this is a situation where using pydicom with its robust DICOM handling and actively maintained DICOM dictionaries might be the optimal solution. I like the idea of saving these as a JSON file, but maybe not the same JSON used as a BIDS sidecar.