`tool_context.actions.skip_summarization = True` makes the agent's response be flagged as coming from user

**Describe the Bug:**

I can create a PR, but I am not sure if this is by design or not...

I am trying to return a tool result directly to the user (without summarization). The result is a User message, not an Assistant/model message. ([CODE](https://github.com/google/adk-python/blob/543f55f5ba78a80633f5e31791d84007baeecd08/src/google/adk/flows/llm_flows/functions.py#L1637))

**Steps to Reproduce:**

Create an agent with the following tool using ADK Python 2.6.3:

```python
from google.adk.tools import ToolContext

def check_password(password: str, tool_context: ToolContext):
tool_context.actions.skip_summarization = True
return "Your password is correct!"
```

**Expected Behavior:**

I would expect the message to be considered a Model/Assistant message (correct me if I am wrong, please)

**Observed Behavior:**

The message was labeled as coming from the user.

**Environment Details:**

- ADK Library Version (pip show google-adk): 2.6.3
- Desktop OS:** [e.g., macOS, Linux, Windows]: Windows
- Python Version (python -V): 3.14.3

**Model Information:**

- Are you using LiteLLM: No
- Which model is being used: e.g., gemini-3.5-flash

**How often has this issue occurred?:**
- Always (100%)