Increase the number of unit-tests for Tensor

Context

Tensor class is at the core of Aidge framework but its tests are very sparse and few core functionnalities are truly tested. Moreover, these tests are not up to the most recent standards using comparison with randomly generated arrays over several iterations.

In the set of MR to improve Tensor, this one aims at improving tests coverage.

Modified files

• Tensor.hpp / Tensor.cpp
• TensorImpl.hpp TensorImpl.cpp
• backend/cpu/TensorImpl.hpp
• Test_TensorImpl.cpp

Detailed major modifications

close #84 (closed)

Add Test_Tensor.cpp file.

Increase cpp coverage by 1.3%

TODO

• Better integration of scalar
  • Add Tensor constructor from scalar
  • adapt Tensor::resize() to scalar Tensor
• Improve unit-tests
  • ↗ coverage
  • + randomized loops
  • + comments
  • better tests structure
• Fix Array to Tensor constructor
  • rightly update DataType and mImpl size
  • remove duplicate code in Array operator= constructors
• Add zeros() member function