[SYCL][NATIVECPU] Port ock compiler internals into sycl native cpu

[coldav]

It is desirable to decouple native cpu from needing to fetch the external oneAPI Construction Kit to allow it to be maintained more smoothly.

To allow a smooth port, much of the files have been copied verbatim, but with a different use of cmake to git in with llvm components. There is a great deal of tidying up that will be required after this port to fit better in with LLVM and prune what is not necessary.

Some documentation has been added, but this will be built upon in future commits.

[aelovikov-intel]

This shouldn't be done by copying (and hence GH PR makes little sense), we need to preserve git history. Please look at how sycl/test-e2e was added.

[coldav]

This shouldn't be done by copying (and hence GH PR makes little sense), we need to preserve git history. Please look at how sycl/test-e2e was added.

I can look into a subtree, I was worried about dragging in the whole of ock for just a few files (also we scrapped history before 2023 when we went open source). I have a limited time-scale so I still think it's useful to look at this PR to see what the final merge might look like - it would probably make sense to make a new PR with the history in it. Also note after all these changes I have made to prune it, it would be almost impossible to subtree merge again.

Note there is some code_formatter fails due to use of undef. I'm not sure what to do with that or if there is a way to disable it or ignore it in the short term.

[aelovikov-intel]

#8782 is how I moved it back then, I didn't use "subtree" but I "filtered" the repo to keep only files we needed.

It was a PR, but I think we temporarily changed merged strategy from "squash" to "merge". Maybe "/merge" can do that now without changing any settings.

[coldav]

okay I am closing this one, so I can create a new one with history, which I'll do next week. Thanks for the feedback.