Part of #430.

Convert packages/scope42-data from a YAML-whole-file parser into a format-agnostic frontmatter extractor driven by a workspace config.

Tasks

• Write FORMAT.md spec: frontmatter fields per type, workspace config schema, extension agnosticism, wikilink semantics. See app/example for reference.
• Define Zod schema for scope42.yaml (item types → paths + idPattern + optional description; global extensions list)
• Trim item Zod schemas in packages/scope42-data/src/model/ to frontmatter fields only (drop description, context, drivers, options, outcome, deciders, comments from the schema)
• Replace YAML-whole-file parser in packages/scope42-data/src/io/workspace.ts with a frontmatter extractor that works on any text file (e.g. gray-matter); do not hard-code Markdown
• Workspace loader reads scope42.yaml, enumerates configured paths, filters by configured extensions, parses frontmatter, infers type from the source path
• Expose parsed items as { id, type, frontmatter, body, filePath } — body is returned as raw text, unparsed
• Drop the current .yml-only assumption; ID derives from filename without extension
• Migrate all app/example into a new directory example with the new format
• Update @scope42/data README (see examples/data-processing for reference)