Multiple output var support

[akapet00]

Resolves #53.

This PR introduces slight changes in API: instead of providing list of functions that will be used to extract relevant features from the simulated traces, the dictionary should be provided instead. Keys should be names of output variables as defined in the model equations, while values should be lists (with the same number of elements) where each element of the list is callable that returns a summary feature. So instead of:

inferencer = Inferencer(dt=dt, model=eqs,
                        input={'I': inp_traces*amp},
                        output={'v': out_traces*mV},
                        features=[lambda x: x[(t > 5) & (t < 10)].mean(),
                                  lambda x: x[(t > 5) & (t < 10)].std(),
                                  lambda x: x.max()],
                        method='exponential_euler',
                        threshold='m > 0.5',
                        refractory='m > 0.5',
                        param_init={'v': 'VT'})

goes:

inferencer = Inferencer(dt=dt, model=eqs,
                        input={'I': inp_traces*amp},
                        output={'v': out_traces*mV},
                        features={'v': [lambda x: x[(t > 5) & (t < 10)].mean(),
                                        lambda x: x[(t > 5) & (t < 10)].std(),
                                        lambda x: x.max()]},
                        method='exponential_euler',
                        threshold='m > 0.5',
                        refractory='m > 0.5',
                        param_init={'v': 'VT'})

This enables the user to simultaneously infer unknown parameters from different output variables.

Additional updates:

• docstrings typos are fixed;
• the relevant examples are aligned with the changes in the API ;
• default level value in some function in Inferencer are now changed, so the user should not have to worry about namespaces even when using a flexible interface, i.e., when using infer_step method manually.

[akapet00]

Awesome, thanks.