[Torch] Do not wrap not tensor inputs

[daniil-lyakhov]

Changes

• NNCF wrapping for Tensor input parameters is skipped when actual parameter is not a tensor

Reason for changes

• To support models which could accept different types as an input (example: YolovV8

def forward(self, x, *args, **kwargs):
    if isinstance(x, dict):  # for cases of training and validating while training.
        return self.loss(x, *args, **kwargs)
    return self.predict(x, *args, **kwargs)

Related tickets

138682

Tests

• tests/torch/test_input_management.py is extended

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 85.71429% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 77.96%. Comparing base (fa1a4ce) to head (3a15424). Report is 2 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff              @@
##           develop    #2649       +/-   ##
============================================
- Coverage    89.84%   77.96%   -11.89%     
============================================
  Files          494      493        -1     
  Lines        45374    45450       +76     
============================================
- Hits         40768    35434     -5334     
- Misses        4606    10016     +5410     

Files	Coverage Δ
nncf/torch/dynamic_graph/io_handling.py	93.81% <85.71%> (+0.69%)	:arrow_up:

... and 108 files with indirect coverage changes

Flag	Coverage Δ
ONNX	?
OPENVINO	?
TENSORFLOW	30.08% <0.00%> (-0.05%)	:arrow_down:
TORCH	66.00% <85.71%> (+0.04%)	:arrow_up:

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
common	88.28% <ø> (-1.26%)	:arrow_down:
torch	93.60% <85.71%> (+0.10%)	:arrow_up:
tensorflow	93.74% <ø> (-0.01%)	:arrow_down:
onnx	0.00% <ø> (-93.08%)	:arrow_down:
openvino	25.67% <ø> (-68.50%)	:arrow_down:
ptq	53.03% <ø> (-32.17%)	:arrow_down: