Use --offload-compress linker option to compress offload sections

See https://www.intel.com/content/www/us/en/developer/articles/technical/sycl-compilation-device-image-compression.html

It is applicable for any SYCL targets. This change results in 28.4% reduction in
shared objects sizes with offload sections on Linux.

Author: oleksandr-pavlyk

dpctl/CMakeLists.txt

@@ -111,6 +111,8 @@ function(build_dpctl_ext _trgt _src _dest)
     Python_add_library(${_trgt} MODULE WITH_SOABI ${_generated_src})
     if (BUILD_DPCTL_EXT_SYCL)
         add_sycl_to_target(TARGET ${_trgt} SOURCES ${_generated_src})
+	target_compile_options(${_trgt} PRIVATE -fno-sycl-id-queries-fit-in-int)
+	target_link_options(${_trgt} PRIVATE -fsycl-device-code-split=per_kernel --offload-compress)
         if(_dpctl_sycl_targets)
             # make fat binary
             target_compile_options(

dpctl/tensor/CMakeLists.txt

@@ -267,7 +267,7 @@ endforeach()
 set(_linker_options "LINKER:${DPCTL_LDFLAGS}")
 foreach(python_module_name ${_py_trgts})
     target_compile_options(${python_module_name} PRIVATE -fno-sycl-id-queries-fit-in-int)
-    target_link_options(${python_module_name} PRIVATE -fsycl-device-code-split=per_kernel)
+    target_link_options(${python_module_name} PRIVATE -fsycl-device-code-split=per_kernel --offload-compress)
     target_include_directories(${python_module_name}
         PRIVATE
         ${CMAKE_CURRENT_SOURCE_DIR}/libtensor/include

dpctl/utils/CMakeLists.txt

@@ -28,7 +28,7 @@ list(APPEND _pybind11_targets ${python_module_name})
 set(_linker_options "LINKER:${DPCTL_LDFLAGS}")
 foreach(python_module_name ${_pybind11_targets})
     target_compile_options(${python_module_name} PRIVATE -fno-sycl-id-queries-fit-in-int)
-    target_link_options(${python_module_name} PRIVATE -fsycl-device-code-split=per_kernel)
+    target_link_options(${python_module_name} PRIVATE -fsycl-device-code-split=per_kernel --offload-compress)
     target_include_directories(${python_module_name}
         PRIVATE
         ${CMAKE_CURRENT_SOURCE_DIR}/libtensor/include