What is an async iterator and async for?

Question presented to candidate: "You already know how a normal for loop works over an iterator. What changes when the next item has to come from the network? Explain what an async iterator is, what async for actually does, and how you would write one."

What a strong answer should cover:

• The protocol: __aiter__ (a plain method that returns the async iterator) and __anext__ (returns an awaitable, usually an async def), ending with StopAsyncIteration instead of StopIteration.
• async for is only legal inside async def; each step awaits __anext__(), so the event loop can run other tasks while the next item is being fetched.
• The easy way to write one is an async generator (async def + yield, PEP 525, Python 3.6); it gets asend, athrow and aclose for free and cannot return a value.
• The helpers: aiter() / anext() builtins (3.10), async comprehensions [x async for x in src] (PEP 530).
• Early exit: after break, an async generator is left suspended and asyncio closes it later; contextlib.aclosing (or await gen.aclose()) makes the cleanup happen at a known point.
• The common errors: async for over a list and a plain for over an async generator are both TypeErrors.

Code / implementation expected: Optional — a small class with __aiter__ / __anext__ or an async generator consumed with async for.