[GPU] Fix reshape in-place optimization with feature padding

[zhanmyz]

Details:

• Fix batch inconsistency error in reshape layer when input tensor has feature dimension padding during in-place optimization.

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

• Symptom: TF_Ssd_Inception_v2_coco model produces incorrect inference results for batch=2 at BoxPredictor_5/Reshape layer [2,24,1,1]→[2,6,1,4].

• Root Cause:

  • BoxPredictor_5/Reshape receives input with feature dimension upper padding (upper_pad=[0,8,0,0]) from upstream convolution layer
  • The has_outer_padding_offset() function in reshape_inst.h only checks input_pad._lower_size[1] != 0 for feature dimension.
  • It misses the upper padding case (input_pad._upper_size[1] != 0), and returns true.
  • In-place optimization reinterprets memory buffer directly, causing reshape to read data from wrong memory offsets due to padding.

• Resolution:

  • Modified reshape_inst.h::has_outer_padding_offset() to also check input_pad._upper_size[1] != 0.
  • Ensures feature dimension with any padding (lower or upper) prevents in-place optimization.

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

openvino/src/plugins/intel_gpu/src/graph/include/reshape_inst.h

Lines 113 to 116 in 849a127

	// Expected a padded input of only batch axis with 'bxxx' format
	if (input_layout.format.dims_order()[0] != 0 ||
	input_pad._lower_size[1] != 0)
	return false;

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

python -m pytest test_ovc_mo.py
-n 2
--tb=native
--env_conf=.automation/env_config.yml
--test_conf=.automation/test_configs/desktop_test_config_gpu_llm.yml
-m "not launch_only_if_manually_specified"
--pregen_irs=models/irs_mapping.csv
--tf_models_version=1.15.2
--modules pipelines/production/tf/light
-k "TF_Ssd_Inception_v2_coco_api_2_True"
--dynamism_type=None
--log-cli-level INFO

Problematic graph

Checklist

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

Tickets:

• CVS-175476