fix(server): honor configured output default across APIs

[earlvanze]

Closes #395.

Threads the server-level --max-output-tokens fallback through the Chat Completions and Completions adapters, as the Responses adapter already does. Explicit request limits still take precedence; absent server configuration retains the 32k fallback.

Validation:

• python3 -m py_compile for changed modules
• git diff --check
• Focused pytest suite is not runnable in this checkout because pytest/runtime dependencies are not installed.