Calling `get_probe_output` multiple times changes result

[tbekolay]

It was discovered in #264 that calling Emulator.get_probe_output multiple times with the same probe would give different results due to the Synapse maintaining state across calls. A test was added to reproduce the bug, but marked as an xfail to not hold up the PR queue (see test_multiple_get_probe_output). It's definitely a bug that should be fixed.

[hunse]

Note that this DOES have an effect even if not using the interface directly (i.e. even if doing everything through nengo_loihi.Simulator). I wasn't sure if it would. It also shows up in HardwareInterface.

Here's an integration test that exhibits the problem (the assertion will fail):

import nengo
import nengo_loihi
import numpy as np

with nengo.Network(seed=0) as net:
    u = nengo.Node(lambda t: np.sin(t))
    e = nengo.Ensemble(100, 1)
    ep = nengo.Probe(e, synapse=0.01)
    nengo.Connection(u, e)

with nengo_loihi.Simulator(net) as sim0:
    sim0.run(1.0)

with nengo_loihi.Simulator(net) as sim1:
    sim1.run(0.5)
    sim1.run(0.5)

assert np.array_equal(sim0.data[ep], sim1.data[ep])