Add Tuning Engines provider catalog

[cerebrixos]

Summary

Adds Tuning Engines as an OpenAI-compatible provider in the Models.dev catalog.

Why this belongs here

Tuning Engines is a governed AI control plane for teams that already use OpenAI-compatible clients, coding agents, and workflow frameworks. The runtime endpoint is OpenAI-compatible, so tools that consume Models.dev can configure it with:

• provider: Tuning Engines
• base URL: https://api.tuningengines.com/v1
• API key env: TUNING_ENGINES_API_KEY

The useful distinction for Models.dev users is that Tuning Engines is not trying to replace the model/client framework. It sits between the app and the selected model so teams can centralize tenant keys, model routing, RBAC, guardrails, audit logs, traces, approvals, and usage/cost controls.

I kept the model list small and representative:

• auto for tenant-configured routing
• common OpenAI/Anthropic/Google/Kimi aliases that a tenant may enable in their catalog

Actual model availability remains controlled by the tenant's Tuning Engines catalog and role permissions.

Validation

• bun install
• bun run validate