feat(crafter): Extract evidence information from input

[javirln]

This PR extends the Evidence crafter to automatically parse JSON evidence files and extract structured metadata as material annotations. This matches with the custom policies documentation which recommends JSON format for evidence materials.

Changes

• Modified evidence.go: Extended the Craft method to intelligently parse evidence files and extract metadata when they follow a specific JSON structure
• Added tryExtractAnnotations helper: Attempts to parse evidence as JSON and extracts id and schema fields as annotations
• Defined customEvidence struct: Represents the expected structure with id, schema (optional), and data fields
• Added annotation constants:
  • chainloop.material.evidence.id - Evidence identifier
  • chainloop.material.evidence.schema - Schema reference URL

Example from an attestation

{
  "annotations": {
    "chainloop.material.cas": true,
    "chainloop.material.evidence.id": "my-awesome-report-id",
    "chainloop.material.name": "material-1761835944639691000",
    "chainloop.material.type": "EVIDENCE"
  },
  "digest": {
    "sha256": "337199c3efb1cf6a579b6f035597ad9a16a178a42a962eb1c32f5f28ce6f770e"
  },
  "name": "report.json"
}