Cannot compile minimal model with one gate

[ligerlac]

While debugging my own code, I noticed that CompiledLogicNet cannot be used to compile a model with just a single logic gate:

llkw = {
    'connections': 'random',
    'implementation': 'python',
    'device': 'cpu'
}
model = torch.nn.Sequential(
    torch.nn.Flatten(),
    LogicLayer(in_dim=2, out_dim=1, **llkw),
    GroupSum(1, tau=1)
)
compiled_model = CompiledLogicNet(
    model=model,
    num_bits=8,
    cpu_compiler='gcc',
    verbose=False            
)
compiled_model.compile(
    save_lib_path='minimal_example.so',
    verbose=False
)

leads to the following error message:

error: use of undeclared identifier 'v0'
        out[0] = (char) (~v0 | v1);
                          ^
error: use of undeclared identifier 'v1'
        out[0] = (char) (~v0 | v1);
                               ^

This is because it produces the following c-code:

void logic_gate_net(char const *inp, char *out) {
	out[0] = (char) (~v0 | v1);
}

where v0 and v1 are not declared.

[ligerlac]

Should be fixed by #21