Add ability to restrict tool arguments to AgentComponent

Problem to solve

Add tool_arguments_binding feature to AgentComponent which will provide a security mechanism to mitigate impact of prompt injection attacks by restricting the scope in which an agent operates. It should overriding tool call arguments at execution time, ensuring that agents cannot be manipulated into accessing resources outside their prescribed data perimeter.

Proposal

Further details

1. Add tool arguments binding to AgentComponent in experimental version
2. Verify the feature
3. Move the feature to v1 version

Links / references