Full support for numpy arrays and lists in Connect

[Silmathoron]

Following a discussion in #1429, I am adding an issue here regarding the support of lists and arrays in Connect.

Is your feature request related to a problem? Please describe.
Currently, many forms of Connect are not supported, or require superfluous arguments:

• list objects are not supported
• using numpy arrays require both inputs to be numpy arrays and the syn_spec argument

import nest
import numpy as np

# create neurons and devices
neuron = nest.Create("iaf_psc_alpha", 10)

vm = nest.Create("voltmeter", 10)

# set weights
weights = np.ones(len(neuron))
delays = np.ones(len(neuron))

# numpy version
arr_vm  = np.array(vm.tolist(), dtype=int)
arr_nrn = np.array(neuron.tolist(), dtype=int)

# works
nest.Connect(arr_vm, arr_nrn, syn_spec={'weight': weights, 'delay': delays, 'synapse_model': 'static_synapse'})

# does not work (but should)
nest.Connect(arr_vm, neuron.tolist(), syn_spec={'weight': weights, 'delay': delays, 'synapse_model': 'static_synapse'})
nest.Connect(vm.tolist(), neuron.tolist(), syn_spec={'weight': weights, 'delay': delays, 'synapse_model': 'static_synapse'})

nest.Connect(arr_vm, arr_nrn)
nest.Connect(arr_vm, neuron, syn_spec={'weight': weights, 'delay': delays, 'synapse_model': 'static_synapse'})
nest.Connect(arr_vm, neuron)
nest.Connect(vm.tolist(), neuron)

Describe the solution you'd like
All examples shown above should be accepted.

Describe alternatives you've considered
We are currently already checking the type of the input in Connect, thus, the solution to automatically convert to NodeCollection if the ids are unique would only require a very minor overhead and no additional code complexity (just a check_input_nodes function that would replace the current test).
I am not sure why syn_spec is currently required, but, to me, when input nodes are not unique, we should be able to switch to the "data connect" without any additional argument...

[stinebuu]

@Silmathoron This is very interesting. Is the reason for

nest.Connect(vm.tolist(), neuron.tolist(), syn_spec={'weight': weights, 'delay': delays, 'synapse_model': 'static_synapse'})

that you want lists to be supported? Because or else you can just use the NodeCollections.

[Silmathoron]

Yes, of course, sorry for the potentially confusing syntax.
All examples are just to show whether the type/combination of types is accepted, so I used .tolist() for simplicity.