data infer-schema fails on top-level JSON arrays (documented example)

[roll]

data infer-schema / infer_data_schema fails on top-level JSON arrays (the documented example).

Environment

• fairspec 0.1.4 (installed from PyPI)
• Python 3.12.3, Linux
• Reproduced in a clean virtualenv against the published wheel.

Repro

cat > users.json <<'JSON'
[
  {"id": 1, "name": "Alice", "email": "alice@example.com"},
  {"id": 2, "name": "Bob",   "email": "bob@example.com"}
]
JSON
fairspec data infer-schema users.json --json

Actual

{ "error": "Invalid descriptor: [\n  {\"id\": 1, ...}\n]\n" }

With --debug the traceback ends in fairspec_metadata/actions/descriptor/parse.py:

def parse_descriptor(text: str) -> Descriptor:
    value = json.loads(text)
    if not isinstance(value, dict):
        raise Error(f"Invalid descriptor: {text}")   # <-- top-level arrays rejected

A single top-level object ({...}) works fine.

Why it's a bug: docs/terminal/data.md documents exactly this case — fairspec data infer-schema users.json where users.json is a top-level array — and shows the expected output as {"type": "array", "items": {...}}. The data-schema path goes through parse_descriptor (descriptor parsing), which is too strict for arbitrary JSON data.

Expected: infer a Data Schema from any valid JSON value, including a top-level array.