[GPU] Update condition of winograd kernel selection

[wilson-seok]

Description of the issue(symptom, root-cause, how it was resolved)

• Winograd kernel selected convolution has wrong accuracy output.
• Padding > 0 would affect to wrong accuracy.
• Added padding check in should_use_winograd_2x3_s1()

The code and line that caused this issue (if it is not changed directly)

• openvino/src/plugins/intel_gpu/src/graph/layout_optimizer.cpp

  Line 487 in 43018d2

  bool should_use_winograd_2x3_s1(const convolution_node& node,

Reproduction step and snapshot (if applicable. Do not attach for customer model)

• $ ./benchmark_app -d GPU.0 -m ~/task/cvs179212/ttybulew.ov.instance-segmentation-bug-001/model.xml -i ~/task/cvs179212/ttybulew.ov.instance-segmentation-bug-001/image.npy -hint none -nstreams 1 -nireq 1 -niter 1

Problematic graph

• I

Checklist

• Is it a proper fix? (not a workaround)
• Did you include test case for this fix, if necessary?
• Did you review existing test that can be extended to cover this scenario? Which test did you review?

Tickets:

• 179212

[byungilm]

Is this difficult to add relevant test-cases?

[wilson-seok]

@byungilm It's hard to create a proper test case about winograd kernel selection. Because the kernel requires byxf format and byfx format selection in get_preferred_format() depends on not only should_use_winograd_2x3_s1() but also +2 depth of convolution_byxf_opt().