Inference adapter for OpenAI-compatible APIs.

Handles wire protocol translation:

• Request serialization (messages, tools, provider routing, multimodal content)
• System role demotion at serialization time (profile-driven)
• Response parsing (faithful field mapping)
• Image generation via /images/generations
• Retry on transient HTTP errors (429, 502, 503, 529)

Model-specific normalization (content fallback, XML tool-call extraction) is handled by Response.Normalizer in the Gateway layer.