Re-implement SYCL backend `parallel_for` to improve bandwidth utilization #210

Job	Run time
CPU,bknd=dpcpp,cmplr=icpx,ubuntu-latest,std=с++17,cfg=release	15m 41s
clang-format	11s
HOST,bknd=tbb,cmplr=icx-cl,windows-latest,std=c++17,cfg=release	18m 6s
FPGA_EMU,bknd=dpcpp,cmplr=icpx,ubuntu-latest,std=с++17,cfg=release	24m 44s
HOST,bknd=omp,cmplr=clang++,macos-latest,std=c++17,cfg=release	4m 29s
HOST,bknd=tbb,cmplr=cl,windows-latest,std=c++17,cfg=release	16m 43s
HOST,bknd=tbb,cmplr=icpx,ubuntu-latest,std=с++17,cfg=release	12m 9s
CPU,bknd=dpcpp,cmplr=icx,windows-latest,std=c++17,cfg=release	12m 17s
HOST,bknd=omp,cmplr=icpx,ubuntu-latest,std=с++17,cfg=release	10m 2s
HOST,bknd=omp,cmplr=g++,ubuntu-latest,std=с++17,cfg=release	6m 11s
HOST,bknd=tbb,cmplr=g++,ubuntu-latest,std=с++17,cfg=release	9m 16s
HOST,bknd=tbb,cmplr=g++,ubuntu-latest,std=с++17,cfg=release	9m 12s
HOST,bknd=tbb,cmplr=clang++,ubuntu-latest,std=с++17,cfg=release	11m 12s
HOST,bknd=omp,cmplr=clang++,ubuntu-latest,std=с++17,cfg=release	8m 53s
HOST,bknd=tbb,cmplr=clang++,ubuntu-latest,std=с++17,cfg=release	11m 24s
HOST,bknd=tbb,cmplr=icpx,ubuntu-latest,std=с++17,cfg=release	11m 53s
HOST,bknd=serial,cmplr=g++,ubuntu-latest,std=с++17,cfg=release	5m 4s
CPU,bknd=dpcpp,cmplr=icpx,ubuntu-latest,std=с++20,cfg=release	27m 10s
HOST,bknd=tbb,cmplr=icpx,ubuntu-latest,std=с++20,cfg=release	18m 0s
HOST,bknd=omp,cmplr=icpx,ubuntu-latest,std=с++20,cfg=release	18m 19s
	4h 10m 56s

Provide feedback