Widen schema:inDefinedTermSet and variableMeasured @type

[smrgeoinfo]

Two schema.org property definitions were narrower than their documented intent, leaving real cases with no valid spelling. Both were found through downstream use in the ADA geochemistry building blocks.

1. definedTerm — schema:inDefinedTermSet

-    type: string
+    anyOf:
+      - type: string
+      - type: object
+        required: ['@id']
+        properties: {'@id': {type: string}}

The class's own union-type policy, stated in its description, already says a DefinedTerm may be a plain string or a full object — but the vocabulary pointer inside it admitted only a string, so {"@id": "…"}, the natural JSON-LD spelling of a reference, was invalid. Both forms are already in use across CDIF profiles and downstream.

2. variableMeasured — @type

     contains:
-      const: schema:PropertyValue
+      anyOf:
+        - const: schema:PropertyValue
+        - const: schema:PropertyValueSpecification

A variableMeasured item is either a measured value or a specification of one — the permitted range, default or unit rather than an observed result. Protocol and method descriptions declare the latter: an ADA TAPP states what a procedure will report before any measurement exists. Pinning @type to PropertyValue alone left that case unspellable.

For reviewers: cdifInstanceVariable separately requires cdi:InstanceVariable in the same array, and that is unchanged. The effective rule for a CDIF instance variable is still "cdi:InstanceVariable AND one of the two schema.org types" — this only widens which schema.org type satisfies the second half. Additional type URIs remain permitted.

Verification

Example validation	134 passed, 8 failed of 142
Same run with changes stashed	134 passed, 8 failed — identical
resolve_schema.py --all	92 schemas, 56 updated, 36 current
Resolver output	no temp-dir $comment stamps, no dangling fragments

Widening these rejects nothing that previously validated — the before/after was measured by stashing, not assumed.

Why the resolved schemas are in this PR

The 56 regenerated resolvedSchema.json files are tracked artifacts, and downstream consumers read the resolved form rather than the source. Committing the source edits alone would leave 61 resolved copies asserting a constraint the source no longer has.

Note that downstream repos fetch these over HTTP from published gh-pages, so the change only reaches them once gh-pages republishes.

🤖 Generated with Claude Code