[PATCH 0/3] x86: fold AVX/AVX2 templates where possible

[Jan Beulich]

This is quite similar to the prior folding of AVX512VL templates
with their AVX512F counterparts, where such exist and where
operand combinations actually allow doing so. Actually the 3rd
patch also folds a few further AVX512 templates which couldn't
be folded earlier on, paralleling the similar AVX2 ones which
are dealt with there.

1: fold certain AVX2 templates into their AVX counterparts
2: drop stray CheckRegSize from VEXTRACT{F,I}32X4
3: also fold remaining multi-vector-size shift insns

Jan