idea: Supports SYCL for Intel GPU

[fandigunawan]

Problem Statement

Supporting Intel GPU is beneficial for large number of users.

Feature Idea

llama.cpp is already support SYCL. SYCL is used by Intel GPU therefore cortex can be used by large number of users that is currently using Intel embedded or discrete GPU.

[qnixsynapse]

Depends on fix for intel/compute-runtime#803 on the recent version of SYCL oneapi base toolkit