Add cases to skip list for test_transformers.py and remove passed cases

[PenghuiCheng]

1. Because XPU didn't support F.scaled_dot_product_attention and torch._scaled_dot_product_efficient_attention with SDPBackend.EFFICIENT_ATTENTION. So Add UT cases to skip list for test_transformers.py.
2. Remove passed cases from skip list.

[Copilot]

Pull Request Overview

Introduces two new tests for memory-efficient attention failure modes on XPU (due to lack of support for the efficient SDP backend) and cleans up the skip list by removing tests that now pass.

• Add imports and two failure-mode tests for large-batch scaled dot-product attention.
• Remove outdated skip entries and prepare skip list for new transformer tests.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
test/xpu/test_transformers_xpu.py	Added imports for F.scaled_dot_product_attention, sdpa_kernel, and two new tests on XPU failure modes.
test/xpu/skip_list_common.py	Removed many outdated skip entries and added a placeholder for skipping the new transformer tests.

Comments suppressed due to low confidence (2)

test/xpu/skip_list_common.py:617

• The new tests test_mem_eff_attention_fail_with_batch_size_geq_65536 and test_mem_eff_attention_fail_with_batch_size_geq_65536_error are only commented out in the skip list. They should be added as active skip entries under test_transformers_xpu.py to prevent unintended failures on XPU.

#        "test_mem_eff_attention_fail_with_batch_size_geq_65536",

test/xpu/test_transformers_xpu.py:56

• [nitpick] Consider adding a short docstring or inline comment explaining the test’s purpose and expected behavior to help future readers understand why this helper is needed.

def _test_mem_eff_attention_fail_with_batch_size_geq_65536(self):