[SYCL] Perform eager initialization on demand

[smaslov-intel]

Under SYCL_EAGER_INIT create as much as possible supplementary data at "warmup"
E2E test in intel/llvm-test-suite#1091
Signed-off-by: Sergey V Maslov [email protected]

[smaslov-intel]

/verify with intel/llvm-test-suite#1091

[smaslov-intel]

The CUDA testing failure is unrelated:

Error response from daemon: failed to create shim task: OCI runtime create failed: error executing hook /usr/bin/nvidia-container-runtime-hook (exit code: 1): unknown
Error: failed to start containers: 8a056352ad40a2c527dc74be0c5266f656b9ee09281a55e554bedfd5e5121565
Error: Docker start fail with exit code 1

[bader]

The CUDA testing failure is unrelated:

Error response from daemon: failed to create shim task: OCI runtime create failed: error executing hook /usr/bin/nvidia-container-runtime-hook (exit code: 1): unknown
Error: failed to start containers: 8a056352ad40a2c527dc74be0c5266f656b9ee09281a55e554bedfd5e5121565
Error: Docker start fail with exit code 1

@tfzhu, I've seen the same message in one of my pre-commits as well. I think CUDA machine is in broken state. Can you fix it, please?

[againull]

LGTM