DataHarmonizer's linkml-datastructure branch doesn't use our validation rules

[turbomam]

Validation rules currently come from the validation_converter sheet and are processed by sheets_and_friends/converters/linkml2dataharmonizer.py

That gets bypassed when we use DataHarmonizer's linkml-datastructure branch, which takes our nmdc_dh.yaml as input

see also https://github.com/cidgoh/DataHarmonizer/issues/290

[turbomam]

“{float} {unit}”

string serialization is interpolation where the things in curly brackets are slots

[turbomam]

full_name: string_serialization: “{given} {family}”

intent is to assemble/serialize a string from smaller value components

means that things with a full_name must also have a given value and a familyvalue

But we have erroneously put what should be MIxS structured patterns into string_serialization slots

[turbomam]

structured pattern: intent is to validate (reuse of smaller validation components)

[sujaypatil96]

On course to make a PR which adds pattern property to the slots in the NMDC schema generated from sheets_and_friends.

[turbomam]

@sujaypatil96 @ssarrafan I see some good developments for this issue in #115, but I don't think I understand how to use it. I have added some discussion about this to the dh_testing README