Add spike support

[akapet00]

Resolves #54.

• added parameter restart to Inferener class. If it is set to true, inferencer object will perform the complete inferencer cycle from the beginning, otherwise it will pass trainer posterior as the proposal from the previous training procedure and it will perform multi-round focused inference.
• added tqdm progress bar instead of manually logging the current round of inference.
• added spike support. The user can now handle spikes as any other output variable (see examples IF_sbi.py and hh_sbi_simple_interface.py):

inferencer = Inferencer(dt=..., model=...,
                        input={'I': input_trace},
                        output={'spikes': spike_times_list},
                        features={'spikes': spike_features_list},
                        ...)

• all examples are updated and ready to use
• minor docstrings fixes are also added

[akapet00]

Merging this to sbi_support now, if any changes are required, I can do it afterwards.