Better behavior with empty InferenceData

[ColCarroll]

Describe the bug
An empty InferenceData object does not save at all, but will act as though it saved something. I think, but haven't checked, that this is because there is a loop over groups, and if there are no groups, we never save anything.

To Reproduce

import arviz as az
import os

filename = 'test.nc'
print(os.path.exists(filename))  # False
data = az.InferenceData()
filename = data.to_netcdf(filename)
print(os.path.exists(filename))  # False
print(filename)  # test.nc

Expected behavior
I would expect test.nc to exist on disk, as an empty netcdf file (or whatever happens when save is called on an empty netcdf file).

[LakshyAAAgrawal]

Hey, is this issue still unresolved. If it is, I would like to take it up. Could someone point me in the right direction please?

[canyon289]

@LakshyAAAgrawal This was closed by #577

Apologies on not closing this issue