Add support for a continuous acquisition interface for AD dets

[jwlodek]

Uses the Circular Buffer plugin.

I've tested that this works with the ContAcqVimbaDetector. Before I continue it would be good if someone could take this for a spin, or have a look-over.

TODO:

• New tests for the cont acq interface
• Modify existing AD factory fixture to toggle between cont acq/not-cont acq
• Update docs to include information on configuring cont acq detector.

Addresses #596

[jwlodek]

The test failures appear to be all tango related, so nothing to do with the changes in the PR...

[jwlodek]

Tested w/ real camera and most recent changes and it all works as expected.

[coretl]

Looks good, a couple of small suggestions

[coretl]

Should we add a drv_cls = ADBaseIO here in case we want to expose the extra PVs the class has? Most of the time I imagine that doesn't matter, but it would probably be useful to have the option...

[thomashopkins32]

Would it also make sense to do this with cb_suffix -> cb_cls?

[thomashopkins32]

Alternatively, the user might want to pass in a custom controller instead which would already have the custom driver and circular buffer classes.

[coretl]

at the moment there is only one circular buffer class in areaDetector, so I think we should hardcode that, while there are many detector drivers that we should probably support

[jwlodek]

Right, since the Circular Buffer plugin is a default plugin that will be included with every driver, and there is only one variant, I think we can leave as is. There are multiple different writers and drivers, hence passing in classes.

[thomashopkins32]

Good with me then

[coretl]

I'm not a fan of wrapping functions as fixtures, not sure it adds anything...

Suggested change

	@pytest.fixture
	def one_shot_trigger_info_factory():
	def generate_one_shot_trigger_info(
	trigger_mode=DetectorTrigger.INTERNAL, livetime=0.8
	):
	if trigger_mode != DetectorTrigger.INTERNAL:
	return TriggerInfo(
	number_of_triggers=1,
	trigger=trigger_mode,
	livetime=livetime,
	deadtime=0.001,
	)
	else:
	return TriggerInfo(
	number_of_triggers=1, trigger=trigger_mode, livetime=livetime
	)
	return generate_one_shot_trigger_info
	def generate_one_shot_trigger_info(
	trigger_mode=DetectorTrigger.INTERNAL, livetime=0.8
	):
	if trigger_mode != DetectorTrigger.INTERNAL:
	return TriggerInfo(
	number_of_triggers=1,
	trigger=trigger_mode,
	livetime=livetime,
	deadtime=0.001,
	)
	else:
	return TriggerInfo(
	number_of_triggers=1, trigger=trigger_mode, livetime=livetime
	)

[coretl]

...

Suggested change

	async def test_cont_acq_controller_invalid_trigger_mode(
	cont_acq_controller: adcore.ADBaseContAcqController, one_shot_trigger_info_factory
	):
	trigger_info = one_shot_trigger_info_factory(
	trigger_mode=DetectorTrigger.CONSTANT_GATE
	)
	async def test_cont_acq_controller_invalid_trigger_mode(
	cont_acq_controller: adcore.ADBaseContAcqController,
	):
	trigger_info = generate_one_shot_trigger_info(
	trigger_mode=DetectorTrigger.CONSTANT_GATE
	)

[thomashopkins32]

Would it also make sense to do this with cb_suffix -> cb_cls?

[thomashopkins32]

Alternatively, the user might want to pass in a custom controller instead which would already have the custom driver and circular buffer classes.