Add agent_tool() examples for common use cases

[sunilp]

What

Add more examples to examples/agent-as-tool/ demonstrating real-world use cases for the three invocation modes:

• Sync: Document classification, sentiment analysis, entity extraction
• Streaming: Long-form research with progress, data pipeline with partial results
• Conversational: Code review with iterative feedback, negotiation between agents

Why

The current example covers the mechanics but does not show domain-specific patterns. More examples help developers understand when to use each mode.

How

1. Pick one use case from the list above
2. Create a new Python file in examples/agent-as-tool/ (e.g., sentiment_analysis.py)
3. Use the agent_tool() wrapper and WorkflowGraph — see examples/agent-as-tool/main.py for patterns
4. Add a brief description to examples/agent-as-tool/README.md

Getting Started

pip install jamjet
cd examples/agent-as-tool
python main.py  # run existing example to understand the pattern

Key files to read:

• sdk/python/jamjet/agent_tool.py — the agent_tool() function
• examples/agent-as-tool/main.py — existing example with all 3 modes