[UR][SYCL] Add support for zeCommandListAppendLaunchKernelWithArguments()

[ldorau]

No description provided.

[ldorau]

Rebased

[ldorau]

@mmichel11 The unit test TEST_F(CommandGraphTest, LinearInOrderQueue) added by you fails with my pull request:
https://github.com/intel/llvm/actions/runs/18773795439/job/53563700631#step:13:4820

[ RUN      ] CommandGraphTest.LinearInOrderQueue
/__w/llvm/llvm/src/sycl/unittests/Extensions/CommandGraph/LinearGraphOptimization.cpp:42: Failure
Expected equality of these values:
  GraphImplTest::NumPartitionsInOrder(Impl)
    Which is: 2
  NumLinearChains
    Which is: 3

/__w/llvm/llvm/src/sycl/unittests/Extensions/CommandGraph/LinearGraphOptimization.cpp:43: Failure
Expected equality of these values:
  GraphImplTest::NumSyncPoints(Impl)
    Which is: 3
  0

[  FAILED  ] CommandGraphTest.LinearInOrderQueue (0 ms)

when I re-apply the patch 24f54ea that was reverted later (d25d6d6). This is the only unit test that fails with my pull request. I have no idea why. Do you know what should be changed to make it pass?

[mmichel11]

@mmichel11 The unit test TEST_F(CommandGraphTest, LinearInOrderQueue) added by you fails with my pull request: https://github.com/intel/llvm/actions/runs/18773795439/job/53563700631#step:13:4820

[ RUN      ] CommandGraphTest.LinearInOrderQueue
/__w/llvm/llvm/src/sycl/unittests/Extensions/CommandGraph/LinearGraphOptimization.cpp:42: Failure
Expected equality of these values:
  GraphImplTest::NumPartitionsInOrder(Impl)
    Which is: 2
  NumLinearChains
    Which is: 3

/__w/llvm/llvm/src/sycl/unittests/Extensions/CommandGraph/LinearGraphOptimization.cpp:43: Failure
Expected equality of these values:
  GraphImplTest::NumSyncPoints(Impl)
    Which is: 3
  0

[  FAILED  ] CommandGraphTest.LinearInOrderQueue (0 ms)

when I re-apply the patch 24f54ea that was reverted later (d25d6d6). This is the only unit test that fails with my pull request. I have no idea why. Do you know what should be changed to make it pass?

This looks to be reported in #20448, and I was able to reproduce the failure on the main SYCL branch with the same error. I confirmed it is resolved by applying #20451, so you may want to try with those changes. It doesn't appear to me that this PR is related.