CPU performance of batched code should be verified

**Is your feature request related to a problem? Please describe.**

Over the years there have been various comments relative to missed or potentially missed CPU performance , e.g. #63, #5395 , #4714. Today the Clang compilers emit failed vectorization warnings from code that "should" vectorize. How relevant these are to realistic calculations is not clear -- various opinions have been expressed but little actual data presented. In view of this and the batched transition it would be worth revisiting conventional CPU profiling of VMC and DMC and identifying any easy/missed opportunities.

**Describe the solution you'd like**
Current code should be checked to not be slower than old releases for CPU molecular and solid state calculations. While not obsessing over every last %, address any major discrepancies and easy portable opportunities for speedup. e.g. Not taking full advantage of vendor libraries.

The issue names no files, tests, or entry points. Begin by identifying the current batched CPU paths for molecular and solid-state VMC and DMC, then compare them with older releases using conventional CPU profiling. Done means major regressions and easy portable speedups are identified and addressed, including missed vendor-library opportunities.

Written by the indexing model from the issue text.