fix(reactive): forward actor options block in ToolCallingLLMCaller._resolve_llm

Port the options-block merging logic from SimpleLLMAgent._resolve_llm()
(stream_router.py, added in PR #11225 / commit b3851693) to the parallel
resolution path ToolCallingLLMCaller._resolve_llm() in tool_caller.py.

Without this fix, running an actor with --skill would ignore the options
block in the actor YAML, causing openai_api_base and openai_api_key to be
silently dropped.  The provider registry then fell back to OPENAI_API_KEY
from the environment, routing requests to api.openai.com instead of the
configured local llama-swap/llama.cpp backend (HTTP 401).

Changes:
- tool_caller.ToolCallingLLMCaller._resolve_llm: read actor_config.options;
  extract openai_api_key -> __api_key_sentinel; forward allowed keys
  (openai_api_base, timeout, top_p, frequency_penalty, presence_penalty);
  reject reserved keys (provider_type, model_id) and unknown keys with
  logger.warning, matching SimpleLLMAgent behaviour exactly.
- features/actor_run_tool_calling.feature: add five BDD regression
  scenarios (section V, tagged @tdd_issue @tdd_issue_11243) covering
  api_base+key forwarding, allowed extra keys, reserved key rejection,
  unknown key rejection, and top-level precedence over options.
- features/steps/actor_run_tool_calling_steps.py: add corresponding step
  definitions for all five scenarios.

All nox quality gates pass (lint, typecheck, unit_tests 15822/0 fail,
integration_tests 1999/0 fail, coverage_report 97%).

ISSUES CLOSED: #11243

features/actor_run_tool_calling.feature

@@ -272,3 +272,36 @@ Feature: Actor run tool-calling via ToolCallingRuntime
     Given a ToolCallingLLMCaller with an actor config without system_prompt
     When invoke is called and the LLM returns a response with encoded tool call names
     Then the LLMResponse contains the decoded tool calls
+
+  # ---------- V: ToolCallingLLMCaller options block forwarding (#11243) ----------
+
+  @tdd_issue @tdd_issue_11243
+  Scenario: ToolCallingLLMCaller forwards openai_api_base and openai_api_key from options block
+    Given a ToolCallingLLMCaller with actor config options containing openai_api_base and openai_api_key
+    When _resolve_llm is called on the ToolCallingLLMCaller with empty tool_schemas
+    Then create_llm was called with openai_api_base forwarded from options
+    And create_llm was called with __api_key_sentinel extracted from options openai_api_key
+
+  @tdd_issue @tdd_issue_11243
+  Scenario: ToolCallingLLMCaller forwards allowed options keys to create_llm
+    Given a ToolCallingLLMCaller with actor config options containing allowed extra keys
+    When _resolve_llm is called on the ToolCallingLLMCaller with empty tool_schemas
+    Then create_llm was called with the allowed options keys forwarded
+
+  @tdd_issue @tdd_issue_11243
+  Scenario: ToolCallingLLMCaller rejects reserved keys in options block with a warning
+    Given a ToolCallingLLMCaller with actor config options containing reserved key provider_type
+    When _resolve_llm is called on the ToolCallingLLMCaller with empty tool_schemas
+    Then create_llm was NOT called with the reserved key provider_type
+
+  @tdd_issue @tdd_issue_11243
+  Scenario: ToolCallingLLMCaller rejects unknown keys in options block with a warning
+    Given a ToolCallingLLMCaller with actor config options containing unknown key foo_bar
+    When _resolve_llm is called on the ToolCallingLLMCaller with empty tool_schemas
+    Then create_llm was NOT called with the unknown key foo_bar
+
+  @tdd_issue @tdd_issue_11243
+  Scenario: ToolCallingLLMCaller top-level config takes precedence over options block
+    Given a ToolCallingLLMCaller with actor config with top-level temperature and options temperature
+    When _resolve_llm is called on the ToolCallingLLMCaller with empty tool_schemas
+    Then create_llm was called with the top-level temperature value