support screen and eyetracker being passed as dict in Experiment.__init__()

[SiQube]

Description of the problem

It would be nice if the screen and eyetracker arguments in the experiment init could be overloaded to also support dictionaries.

Let's take this dict as an example:

dict_definition = {
    'screen': {
        'height_cm': 30.2,
        'height_px': 1024,
        'width_cm': 38.2,
        'width_px': 1280,
        'distance_cm': 60,
        'origin': 'center',
    },
    'eyetracker': {
        'sampling_rate': 2000,
        'model': 'EyeLink II',
        'vendor': EyeLink',
    },
}

Trying to initialize a DatasetDefinition with DatasetDefinition(**dict_definition) wont' work as screen and eyetracker are just dicts and not the initialized class objects.

#914 implemented a workaround in its DatsetDefinition.from_yaml() method but it would be more elegant to solve this in the Experiment.__init__() itself.

Description of a solution

The screen and eyetracker arguments in Experiment.__init__() should be overloaded to also support dictionaries.
If a dictionary is passed, a Screen or EyeTracker object should be initialized on the fly.

[dkrako]

Isn't the experiment key in the dictionary redundant?

[dkrako]

If this is implemented, this has to be implemented for Screen and EyeTracker as well

EDIT: although probably self.screen = Screen(**kwargs['screen']) could work as well in the Experiment.__init__()

[SiQube]

Isn't the experiment key in the dictionary redundant?

yes -- sorry -- this was just my "copy/paste" from a nested dictionary dataset definition.

[SiQube]

this would also be in line with #914 because in #914 we write each experiment/screen/eyetracker etc. as a dictionary/yaml file. hence it would be great to have this kind of parsing