fix: Fix missing dynamically loaded tools in SkillToolset during the same invocation

Currently, BaseToolset caches its tools per invocation_id. Because SkillToolset dynamically resolves additional tools from the state when a skill is loaded, the cache prevents new tools from being picked up in the same invocation right after a load_skill call. This change sets `_use_invocation_cache = False` in SkillToolset so that it correctly re-evaluates the state-dependent tools at each step of the LLM generation loop within an invocation, preventing "Tool not found" errors.

PiperOrigin-RevId: 914997555

Author: google-genai-bot

src/google/adk/tools/skill_toolset.py

@@ -895,6 +895,8 @@ def __init__(
       script_timeout: Timeout in seconds for shell script execution via
         subprocess.run. Defaults to 300 seconds. Does not apply to Python
         scripts executed via exec().
+      additional_tools: Optional list of `BaseTool` or `BaseToolset` instances
+        to be made available to the agent when certain skills are activated.
     """
     super().__init__()
 
@@ -911,6 +913,8 @@ def __init__(
     self._registry = registry
     self._code_executor = code_executor
     self._script_timeout = script_timeout
+    # Needed for mid-turn reloading of skill tools.
+    self._use_invocation_cache = False
     self._invocation_cache: dict[
         str,
         dict[str, models.Skill | asyncio.Future[models.Skill | None] | None],