Refactor BoardObserver and LEDState

[morgenmuesli]

The current LedStateDetector and BpardObserver combines Data and Logic in one class. This can be a bit confusing for later development.

A better approach would be to split those classes and their Functions on smaller batches to provide readability and modularity.

current

Note: The class diagram is shortened

classDiagram
class StateDetector{
 - currentOrientation: np.array
 - detect_current_state()
 + on_change()

}

class BoardObserver{
 + leds: List[StateDetector]
 - brightness: deque(maxlen=30)
 + check(frame, roid:List, on_change:Callback, *args, **kwargs)

}

class LedStateDetector{
 // led object 
 + id: int
 + name: str
 + is_on: Bool
 + ...

 - state_change()
 + detect()

}

Problems:

• LedStatesDetector is initialised for every led. This is unperformed and consumes much more memory than needed.
• LedStateDetector stores the current LedState for an individual led seperatly instead of using the already existing led_State dataclass
• LedStateDetector stores led.id and led.name this is redundant since the name is the identifier
• StateDetector checks invalide rois which should be checkt by the get_led_roi function.

Possible Solutions

• split LedStateDetector in static function and Dataobjects
• split StateDetector.detect_current_states() in smaller functions
• rename overloaded "detector" names