add AsyncToolset

[longcw]

AsyncToolset lets tools run in the background while the voice agent keeps talking. Example

class TravelToolset(AsyncToolset):
    @function_tool
    async def book_flight(self, ctx: AsyncRunContext, origin: str, dest: str) -> str:
        await ctx.update(f"Searching flights from {origin} to {dest}...")
        # agent replies to user that it's searching the flight

        flights = await search_flights(origin, dest)
        await ctx.update(f"Found {len(flights)} options, booking the best one...")
        # agent will update to user when it's idle

        booking = await confirm_booking(flights[0])
        return f"Booked! Confirmation: {booking.id}"
        # final result will also be updated when both user and agent is idle

agent = Agent(tools=[TravelToolset(id="travel")], instructions="...")

How it works:

• Progress updates: Long-running tools can push intermediate status messages (e.g. "Found 3 flights, confirming the best one...") into the conversation via ctx.update(). The tool updates and results are injected as tool outputs immediately when updates are created.

• Delivery timing: Reply is generated when the agent is idle. Multiple updates from concurrent tools are merged into a single reply to avoid back-to-back replies. If the agent already replied after update injection (e.g. the user asked about it and the LLM answered from context), the tool reply is skipped.

• Duplicate prevention: To avoid calling the same async tool multiple times for the same goal, by default it rejects the duplicate and lets the LLM re-call with explicit confirmation, cancel the existing task, or do nothing.

[theomonnom]

lgtm! pending and update could be merged