Two code paths in the reactive actor run pipeline silently discarded the
options: block from v3 actor YAML, preventing custom OpenAI-compatible
backends (llama.cpp, Ollama, etc.) from being used.

Fix 1 — ReactiveConfigParser._build_from_v3() (config_parser.py):
The options dict was never copied into agent_config. Added a guard
matching the existing pattern for env_vars/response_format/memory:
if options is a non-empty dict, propagate it as agent_config['options'].

Fix 2 — SimpleLLMAgent._resolve_llm() (stream_router.py):
After building llm_kwargs from the fixed keys (temperature, max_tokens,
max_retries), iterate over options items and insert each key that is not
already present. Explicit top-level keys take precedence over duplicates
inside options, which is the least-surprising behaviour.

No changes to registry.create_llm() are needed — it already accepts
**kwargs and passes them through to ChatOpenAI(...).

Tests: 4 new Behave scenarios across actor_v3_schema.feature and
consolidated_routing.feature verify options forwarding end-to-end
through both the parser and the LLM resolver.

ISSUES CLOSED: #11223

@@ -1382,0 +1389,4 @@
+  # M5: actor without options block is unaffected (#11223).
+  @tdd_issue @tdd_issue_11223
+  Scenario: SimpleLLMAgent without options block calls LLM constructor without extra kwargs
+    Given a stream router llm agent without options block

@@ -229,0 +240,4 @@
+        options = dict(self.config.get("options") or {})
+        if "openai_api_key" in options:
+            llm_kwargs["__api_key_sentinel"] = options.pop("openai_api_key")
+        # Ensure sensitive/reserved ChatOpenAI constructor params cannot