[Fix] Dynamically set output Tensor feature of `Identity_Op` based on input Tensor

This solves aidge#234 (closed).

After this fix, the script provided by aidge#234 (closed) produce a clear error:

Assertion failed: exists(key) in /local/is154584/ob222806/aidge/aidge/install/include/aidge/utils/Registrar.hpp:79
missing or invalid registrar key: [Float64, Default, [(1, 1), (1, 1), (2, 2), (8, 8)], Float32, Default, [(0, 0)], Float32, Default, [(0, 0)], Float32, Default, [(4, 4)]], [Float32, Default, [(1, 1), (1, 1), (3, 3), (5, 5)]] for registrable object N5Aidge16OperatorImpl_cpuINS_9Resize_OpEFvPKvRKSt6vectorImSaImEES8_NS_13Interpolation24CoordinateTransformationENS9_4ModeENS_13PadBorderTypeEPvEFvvEEE
Did you include/import the corresponding module?
If so, it is possible that the object is not yet supported.

Meaning there is no implementation for Resize operator which takes Float64 for the first input and Float32 for the other inputs, because the model input is in Float64!