Feat/operator squeeze unsqueeze

Context

see issue : aidge_core#65 (closed)

Refactored tests for onnx

• Test_import_export can now easily run multiple networks
• node register has been completed

replaced all prints by logging

Fixed onnx node opset retrieval

Done in 190c2d9f

Previously the opset of the model was retrieved with model.ir_version, which is .... I do not know what it is but this is named intermediate representation and not opset.

Opsets have to be retrieved via model.opset_import. Its a list [opset] (/protoobject) that contains :

1. a domain name
2. an associated opset

Each node is attached to a domain name which bind them to a given opset.

This allos to use multiple opsets in a given model.

More on that here : http://onnx.ai/sklearn-onnx/auto_tutorial/plot_cbegin_opset.html

aidge_onnx/node_export/aidge_converters/add.py

@@ -21,7 +21,7 @@ def export_add(
     node_outputs_name,
     opset:int = None,
     verbose: bool = False,
-    **kwargs) -> List[onnx.helper.NodeProto]:
+    **kwargs) -> List[helper.NodeProto]:
 
     aidge_operator = aidge_node.get_operator()