Add backend property to experiments

[chriseclectic]

Summary

• Add backend as an optional __init__ kwarg for all experiments to allow setting the backend at initialization. The backand can also be set and retrieved from the experiment object after construction using the BaseExperiment.backend property and setter.
• When using the backend kwarg of BaseExperiment.run to specify a backend this will temporarily override any currently set backends for that single execution.
• Added a _set_backend method to BaseExperiment that is called when a backend is set via initalization or the backend setter. This can be overridden in experiment subclasses if required. For example this could be used to extract any needed configuration or properties from the specified backend, or to update experiment options of configuration based on the backend.

Details and comments

[nkanazawa1989]

This looks almost good to go. Probably future support of backend in the run method is worth discussing.

[nkanazawa1989]

I prefer backend to be always in the second position in the constructor because qubits and backend are common args among experiments (this cannot come before qubits since optional) though this will become a breaking change.

[chriseclectic]

I wanted to keep it as optional arg, because later when we add experiment settings we want to be able to reconstruct all experiments from settings without a backend (since backends cannot be serialized/deserialized as JOSN, only converted to string names). To be able to load an experiment from settings you will likely have to load it without the backend, and then reset the backend after

[nkanazawa1989]

I think this shouldn't be a problem because json is dict and other options are passed as kwargs. We can pop string backend if we want. I think keeping backend an optional arg is fine, but if we move the position, we can also write like

my_experiment = Exp([0, 1], my_backend)

I'm familiar with this argument pattern, e.g. transpile, assemble, schedule, etc...

[chriseclectic]

The above only works for experiments with only 1 positional arg, which is only a couple of the current ones in the library. But I will move to the first kwarg arg.

[nkanazawa1989]

Thanks. I think eventually we can change other options to be optional with good default values (i.e. qccode doesn't always take scan values).

[chriseclectic]

Linting is failing for fine_amplitude_cal tests in a way I can't figure out. When printing the dict passed to transpile that is causing this error it doesn't even have the "gates" key which the linter is complaining about.

[nkanazawa1989]

Suggested change

	self.experiment_options.gate = gate
	self.set_experiment_options(gate=gate)

This solved the error.

[nkanazawa1989]

Probably we need a setter method to avoid future lint errors. This lint error is really hard to find...

@experiment_options.setter
self experiment_options(**fields):
    raise QiskitError("Use set_experiment_options instead.")

[nkanazawa1989]

Sorry this doesn't work. We need to typecast Options to something like FrozenOptions.