doc: add Intel MKL compilation for building library with server example

[xiangyang-95]

This pull request adds steps to build library with server example with Intel MKL backend.
This will also add support for AVX_VNNI hardware instruction sets in Intel CPU.

[vansangpfiev]

Thanks for your PR. We will review it soon.
Seems like the PR has some conflicts, could you please help to fix them?

[xiangyang-95]

@vansangpfiev I have resolved the conflict. Feel free to let me know if need any info or modifications.

[xiangyang-95]

Hi @hiento09, may I know are there any changes needed? I am also planned to contribute with a version running on SYCL as well

[dan-menlo]

@xiangyang-95 FYI, we will be working on Intel in a couple of sprints - thank you for the patience as we get things in order 🙏

[xiangyang-95]

Hi @dan-homebrew, I am actually from Intel team. Please let me know if anything I can help.

[dan-menlo]

Hi @dan-homebrew, I am actually from Intel team. Please let me know if anything I can help.

Hey yes I saw! We also spoke with a few of your colleagues.

On our part, we need to take a bit of time to refactor our implementation of cortex.llama.cpp in Jan:

Jan

• We will allow users to download multiple versions of llama.cpp (e.g. compiled with sycl, avx2)
• We will have a UI which allows user to select which version of llama.cpp they want to use
• We will pre-bundle several versions of llama.cpp, and make additional ones (e.g. sycl available for pull)

Cortex.cpp

• For cortex.cpp, most of the work is right now infrastructural as we bootstrap the CLI and basic functions
• We are implementing cortex engines install, which right now relies on hardware auto-detection
• cortex engines install llama.cpp will need to expand to support the different llama.cpp variants