Add semantix-ai under Audit

[labrat-akhona]

Adds semantix-ai under Audit.

semantix-ai is an open-source (MIT) Python library for validating LLM outputs against plain-English semantic intents using a local quantized NLI model (~25 MB ONNX, ~15 ms per call, deterministic, no API key).

Why it fits Audit:

• Every validation produces a hash-chained JSON-LD Semantic Certificate. Modifying any entry breaks every subsequent hash — tamper-evident by construction.
• CLI semantix verify audit.jsonl re-computes the chain and reports entry counts, pass/fail rates, and top intents.
• Compliance-aligned framing (POPIA §71, GDPR Art. 44, EU AI Act high-risk-system obligations) baked into the design, not retrofitted.
• Drop-in integrations for DSPy, LangChain, Pydantic AI, Guardrails, Instructor, and MCP.

Sits naturally alongside AIR Blackbox (tamper-evident audit chains, trust layer for agent frameworks) and glassalpha.

• Repo: https://github.com/labrat-akhona/semantix-ai
• PyPI: https://pypi.org/project/semantix-ai/
• Docs: https://labrat-akhona.github.io/semantix-ai/
• License: MIT