feat: enrich /v1/models with context_length, pricing, and max_output_tokens

[sanwzzz]

Summary

• Enhance /v1/models to optionally include: context_length, max_output_tokens, pricing.
• Metadata source priority: pricingService (LiteLLM pricing metadata) -> embedded fallback defaults.
• Keep backwards compatibility: fields are omitted when unknown.

Why

• Clients like Hermes rely on these fields to display correct context window and pricing.

Implementation notes

• Added internal/pkg/modelmetadata with embedded default model metadata JSON for stable tests (no cwd dependency).
• Extended pricing_service parsing to support max_input_tokens/max_output_tokens/max_tokens.

Tests

• go test -tags=unit ./internal/handler ./internal/service -count=1

Closes #1777

[github-actions]

All contributors have signed the CLA. ✅
_{Posted by the CLA Assistant Lite bot.}

[sanwzzz]

I have read the CLA Document and I hereby sign the CLA

[sanwzzz]

recheck