Add instance_norm op and test.

[asenina]

Changes

The torch.nn.InstanceNorm3d layer is wrapping by NNCF.

Reason for changes

This is necessary for correct pruning of models with conv3d and InstanceNorm3d

Related tickets

#77188

Tests

tests/torch/pruning/test_model_pruning_analysis.py

[asenina]

Jenkins please retry a build

[vshampor]

Jenkins please retry a build

[asenina]

Jenkins please retry a build

[openvino-nncf-ci]

Can one of the admins verify this patch?

[codecov[bot]]

Codecov Report

Merging #1089 (2ffddf5) into develop (b6910ea) will decrease coverage by 3.91%. The diff coverage is 100.00%.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1089      +/-   ##
===========================================
- Coverage    89.86%   85.96%   -3.91%     
===========================================
  Files          485      485              
  Lines        44491    43777     -714     
===========================================
- Hits         39981    37632    -2349     
- Misses        4510     6145    +1635     

Flag	Coverage Δ
COMMON	15.81% <0.00%> (+0.06%)	:arrow_up:
ONNX	?
OPENVINO	38.66% <60.60%> (+0.03%)	:arrow_up:
TENSORFLOW	30.01% <0.00%> (-0.04%)	:arrow_down:
TORCH	62.74% <100.00%> (+0.02%)	:arrow_up:

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

Files	Coverage Δ
nncf/torch/graph/operator_metatypes.py	99.12% <100.00%> (+<0.01%)	:arrow_up:
nncf/torch/graph/pattern_operations.py	100.00% <100.00%> (ø)
nncf/torch/layers.py	95.05% <100.00%> (+0.23%)	:arrow_up:
nncf/torch/pruning/operations.py	75.85% <100.00%> (+0.05%)	:arrow_up:

... and 56 files with indirect coverage changes

[vshampor]

@daniil-lyakhov the pruning tests are failing, please see what's wrong and advise, or directly push additional fixing commits into this PR.

[daniil-lyakhov]

@vshampor, I tried to find out reasons, but after an hour I think one need to do a deep dive into pruning algo again to solve this. I have no clue why references have changed after LayerNorms presentation, they should not have affected. I suggest to close this PR, otherwise, I feel like I need some approve to spend a day to understand the root cause