feat: add ToolSimulator for tool response simulation

[ybdarrenwang]

Description

Introduces ToolSimulator framework for simulating realistic tool responses during agent evaluation without calling production APIs. Enables systematic testing of agents with function tools through LLM-powered dynamic simulation.

Key capabilities:

• Dynamic (LLM-powered) simulated response generation
• Shared state management across multiple tools via share_state_id for stateful testing scenarios
• Decorator-based registration for function tools
• Integration with Strands Evals workflow including Experiment, Case, and evaluators
• Multi-agent support with tool simulation across sub-agents (agent-as-tool pattern)

Design principles:

• Centralized registry for tool management and state tracking
• Context-aware response generation using initial state descriptions and conversation history
• Seamless integration with existing Strands tool decorator patterns
• Comprehensive unit test coverage for all simulation modes

Related Issues

#93

Documentation PR

strands-agents/docs#500

Type of Change

New feature

Testing

• I ran hatch run prepare

Checklist

• I have read the CONTRIBUTING document
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.