profile: new command to extract and infer metadata from a dataset, given an optional spec file, and create a .qsv file

[jqnatividad]

Now that DCAT-US v3 has reached a recommendation snapshot status -

https://doi-do.github.io/dcat-us/

The metadata command will extract/infer the metadata and save it to a .metadata file.

If a spec file is given, it will use it to inform metadata inferencing (the default spec file will be based on DCAT-US v3 profile).

When uploading to CKAN, the uploader (qsv pro, DP++) will use the .metadata file to prepopulate metadata fields and set the package to DRAFT mode so the metadata can be curated by the Data Publisher.

The .metadata file will also be used to create an expanded data dictionary for the CKAN resource.

[jqnatividad]

For an extended discussion about the data dictionary format, see DOI-DO/dcat-us#138

[jqnatividad]

Call this command profile instead. This is consistent with the verb convention for command names.

Also, the file it produces uses a .profile extension instead of .metadata.

In addition, the DCAT3 spec file will be able to call either luau or python scripts to compute and infer metadata.

[jqnatividad]

Because of the interactive nature of profiling a dataset, the need to use LLMs and other web services, this will need to be done in qsv pro.

[jqnatividad]

But we can still do one-pass DCAT 3 profile creation, once we replace dynfmt with minijinja #1374