Security: Sophia Inspector Prompt Injection Protection

[MichaelSovereign]

Summary

This PR fixes a critical prompt injection vulnerability in the Sophia attestation inspector.

Changes

1. Input Sanitization: Added to clean user-controlled fields (, , , etc.) before they are merged into the LLM prompt. This prevents attackers from hijacking the LLM's instructions via malicious miner names or device strings.
2. Structural Integrity: Replaced indented JSON with compact separators and stricter truncation to ensure the prompt remains within safe token limits and maintains clear instruction boundaries.
3. Defense Against Hijacking: Neutralizes attempts to trick Sophia Elya into auto-approving fraudulent hardware fingerprints.

Closes #7448