[upd] benchmark

Context

As stated in !407 (merged) coming steps, some operators are not compatible with the current benchmark configuration description system.

configuration file format [should be updated] to accept specific values for input Tensor. This is ESSENTIAL to test Operators such as Slice whom inputs like axis or start behave like attributes and cannot be chosen at random

The MR aims at solving this.

Main changes

• upd: change configuration order, remove 'main_parameter' that was redundant, change 'input_shapes' for 'input_properties' and add values to 'input_properties'
• add: management of pre-set values in configuration -> see example with reshape.json
• move configuration management to manage_config.py
• add: many checks for configuration
• start the process of removing many functions from the main benchmark
• Tensor values set in benchmark instead of ONNX model generator to set the ground for a function that would save these Tensors
• add: function in manage_config.py to create config file templates for any ONNX operator

Next updates

• use the automatic test system from configuration of benchmark with unit-tests of Aidge modules
• adapt configuration to test Aidge-only operators (those not available in ONNX)