Initial working python registrar.

This MR aims at allowing to define and register a python implementation which is callable by Aidge.

import aidge_core
import aidge_backend_cpu

class randomImpl(aidge_core.OperatorImpl):
    def __init__(self, op: aidge_core.Operator):
        print("Crating a random Impl !")
        aidge_core.OperatorImpl.__init__(self, op) # Recquired to avoid type error !
        self.op = op # Reference to the operator
    def forward(self):
        print("Random Impl called !")

def createConvImpl(op: aidge_core.Operator):
    print("Pybind registrar call !")
    opImpl = randomImpl(op)
    return opImpl


# REGISTER NEW IMPLEMENTATION
aidge_core.register_ConvOp2D("cpu", createConvImpl)

conv = aidge_core.Conv2D(2,2,[1,1], name="Conv0")
conv.get_operator().set_backend("cpu")
input_node = aidge_core.Producer(aidge_core.Tensor(np.random.randn(1, 2, 2, 2).astype(np.float32)), name="data")

model = aidge_core.sequential([conv])

model.set_backend("cpu")
input_node.get_operator().set_backend("cpu")
input_node.get_operator().set_datatype(aidge_core.DataType.Float32)
input_node.add_child(model)

# Create SCHEDULER
scheduler = aidge_core.SequentialScheduler(model)

# Run inference !
scheduler.forward(verbose=True)