Fix ReduceMax output shape correctness on CPU and GPU.

[ShreyasN707]

Details:

• Fixed an issue where ReduceMax operations were producing incorrect output shapes (e.g., [1, 960] instead of [1]) on both CPU and GPU plugins.
• CPU Plugin: Enabled ConvertReduceMaxToPooling transformation in src/plugins/intel_cpu/src/transformations/transformation_pipeline.cpp. This forces ReduceMax to be converted to - MaxPool, which correctly handles shape inference and keep_dims, avoiding the issue where the operation was dropped or mishandled.
• GPU Plugin: Disabled ConvertReduceToReshape transformation in src/plugins/intel_gpu/src/plugin/transformations_pipeline.cpp. This prevents the plugin from incorrectly optimizing "redundant" reductions into Reshapes, ensuring the ReduceMax operation is preserved.

Tickets:
N/A
@ShreyasN707 Can you please update as much as possbile per following template?

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

• A
• B

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

• intel_gpu/src/… (or func name)

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

• $ benchmark_app …
• ...

Problematic graph

• If it is related to graph, please describe the graph
• If you need to show captured graph, attach on the ticket

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:

• N/A

[mlukasze]

do not modify .gitignore of the project with your local files, please.
to be corrected

[ShreyasN707]

do not modify .gitignore of the project with your local files, please. to be corrected

Ohh i byemistakely pushed that too. Will fix it!

[ShreyasN707]

@isanghao @peterchen-intel please review my PR.

[peterchen-intel]

Suggested change

	# Local CI/CD Check Script for OpenVINO