FIX: warn instead of raising on reasoning-model token truncation

[romanlutz]

Description

Reasoning models spend completion tokens on hidden reasoning before emitting any visible output. When max_completion_tokens is set too low, the API can return finish_reason="length" with empty visible content. OpenAIChatTarget._validate_response treated that empty content as a hard failure and raised EmptyResponseException (surfaced with a misleading "Status Code: 204"), which also triggered the full retry loop even though retrying with the same token budget cannot succeed.

Since a low token budget can be a deliberate configuration choice, this changes the behavior to warn rather than raise:

• On any finish_reason="length", log a warning explaining that reasoning models consume tokens on hidden reasoning and that max_completion_tokens may need to be increased.
• When the truncation left no visible content, return a graceful empty response (error="empty") so the run continues instead of raising and retrying.
• Responses that are empty for other reasons still raise EmptyResponseException as before.

The final stored result for the empty case is unchanged from today (an error="empty" piece). The difference is no exception, no retry storm, and a clear warning pointing at the token limit.

Tests and Documentation

• Updated test_validate_response_success_length to assert the truncation warning is emitted when content is present.
• Added test_validate_response_length_empty_returns_empty_response asserting that a truncated-and-empty response returns a Message with error="empty" and raises nothing.
• Full tests/unit/prompt_target/target/test_openai_chat_target.py suite passes (101 tests).

No documentation changes needed. JupyText not run (no doc or notebook changes).