Scan documents and infer the frontmatter schema.

For each field found across the workspace, displays:
- Field name (dot-notation for nested fields)
- Type distribution with percentages
- Coverage: how many documents contain this field
- Distinct: count of distinct enumerable values
- Value distribution for low-cardinality fields (omitted when more
  than 100 distinct enumerable values)

Only enumerable values are listed: null, booleans, numbers, and
strings made up of [A-Za-z0-9_.-/]. Free-text strings (titles,
URLs with `:`, prose) are counted in coverage and types but do not
appear in the values listing or distinct count.

Use --filter to restrict analysis to a subset of documents (e.g.,
only posts, only externals). Use --field to drill into a specific
field. The --project / --add-fields flags from `find` and `tree` do
not apply here -- schema operates on raw frontmatter.
