Merged requested to merge kvamme62/sse2 into master
Cherry-picked from PR 390676, 391026, 391082, 391600 in our internal repository.
Use SSE2 intrinsics to scan min/max values in an array, to speed up processing.
Also add BuildNoHistogram to the available finalize modes. This is primarily meant for performance testing. It will disable both statistics and histogram. Compare timings with that of BuildFull. Do not use in production. There are caveats. Such as, need to change the low resolution algorithm when BuildNoHistogram is active.
Example command-line test
zgycopyc -i src.zgy -o dst.zgy --finalize nohistogram --algorithm LowPass,Average