Refactor anatomy.cpp for USM and in_order queue

[MikeL270]

1. Updated anatomy.cpp to use malloc_device. We did not use malloc_shared in order to demonstrate the in_order behavior.
2. Modified accompanying description to reference the updated code and clarify that the command group scope is created implicitly because of the parallel_for queue shortcut.

Thank you for taking the time to review my pr and do not hesitate to reach out for clarification if needed.

[CLAassistant]

All committers have signed the CLA.

[gmlueck]

Please retarget this PR to the "main" branch. Our policy is to make all changes on the "main" branch unless the change is somehow not applicable to future versions of the spec. In this case, the change clearly is applicable to future versions.

Once the PR is merged, I will cherry-pick it to the "sycl-2020" branch if the WG approves it as a SYCL 2020 clarification (which I expect would be the case for this PR).

[MikeL270]

Please retarget this PR to the "main" branch. Our policy is to make all changes on the "main" branch unless the change is somehow not applicable to future versions of the spec. In this case, the change clearly is applicable to future versions.

Once the PR is merged, I will cherry-pick it to the "sycl-2020" branch if the WG approves it as a SYCL 2020 clarification (which I expect would be the case for this PR).

Done! I will work on addressing the rest of your feedback shortly. Thank you for the guidance!

[gmlueck]

Thanks, this looks much improved! Just a few comments below.