Add outlines to AI and Agents (Data Layer)

[Oliver-Liu99]

Adds outlines, a library for constrained decoding of LLM outputs, under AI and Agents → Data Layer.

Why it qualifies (all quality requirements met):

• Python-first: 99.7% Python
• Active: commits within the last weeks
• Stable: production-ready, widely used across the structured-generation ecosystem
• Documented: README with multiple real-world examples (classification, extraction, parsing, templating)
• Established: repository ~2 years old with consistent activity

Acceptance criteria — Industry Standard for its niche:
13.9k+ stars and the de-facto tool for guaranteed-valid structured LLM output.

Uniqueness (not a duplicate of instructor):
instructor validates/parses output after generation via Pydantic and retries on failure. outlines enforces validity during token generation through JSON-schema / regex / context-free-grammar constrained decoding, so invalid output is structurally impossible rather than caught afterward. No constrained-decoding library currently exists in this list.