docs: clarify how tools work in Atomic Agents (no tools=[...] parameter) and add a dedicated guide

[KennyVaneetvelde]

Came up again in Discord — beginners coming from frameworks like LangChain/CrewAI/PydanticAI expect to be able to do something like agent = Agent(tools=[calculator]) and are confused when they can't find that in Atomic Agents. The current docs don't explicitly call out that:

• There is no tools=[...] parameter, and that's intentional.
• Tools are just atomic components with an input & output schema, and YOU decide when to call them.
• There are basically two ways to use them in practice:
  1. Call directly when you already know which tool you need at that point in the flow (deterministic, fast, cheap, debuggable).
  2. Use a "choice/decision agent" whose output_schema is a Union of possible tool input schemas — match on the type and run the corresponding tool.
• The deep-research example (deep_research/agents/choice_agent.py) is a real reference for pattern unexpected character after line continuation characte #2.

Action items:

• Add a dedicated "Using Tools" guide explaining both patterns with code snippets.
• Cross-link from the orchestration guide and the README.
• Mention the philosophy upfront so people coming from other frameworks know what to expect (and why).