Simulated AreaDetector still in Acquire state after collection when yield from trigger_and_read(detector, *devices)

[DiamondJoseph]

Stacktrace:
bar.txt

When calling trigger_and_read rapidly on the SimulatedAreaDetector, backed by the containerised IOC

ophyd-async --version
0.9.0a1

Steps To Reproduce

Steps to reproduce the behavior:

1. Create a plan that loops over trigger_and_read with a the ADsim
2. Observe that it occasionally (~0.5% of the time?) is still in the Acquire state

Acceptance Criteria

• The ADSim can reliably be used on scans that collection many, many points without failure.

[DiamondJoseph]

This was observed by @DiamondRC who can provide the plan if behaviour cannot be replicated

[coretl]

This looks like a race condition we fixed in Malcolm, but didn't apply here.
Malcolm:
https://github.com/DiamondLightSource/pymalcolm/blob/264ec4a319092a545853f41f89ddf1d21da56504/malcolm/modules/ca/parts/caactionpart.py#L98-L122
Ophyd-async:

ophyd-async/src/ophyd_async/epics/adcore/_core_logic.py

Lines 101 to 107 in c781e27

	await status
	state = await driver.detector_state.get_value()
	if state not in good_states:
	raise ValueError(
	f"Final detector state {state.value} not in valid end "
	f"states: {good_states}"
	)

So if we changed to:

await status
state = None
try:
    async for state in observe_value(driver.detector_state, done_timeout=STATUS_TIMEOUT):
        if state in good_states:
            return
except asyncio.TimeoutError:
    raise ValueError(
        f"Final detector state {state.value} not in valid end "
        f"states: {good_states}"
    )

or something similar that should fix it...