AffineTransformで`kOutputDimensions`が4で割り切れない場合でもAVX2やAVX-512を使えるようにした。 #1066

Job	Run time
build-android (YANEURAOU_ENGINE_NNUE)	1m 53s
build-android (YANEURAOU_ENGINE_KPPT)	1m 56s
build-android (YANEURAOU_ENGINE_KPP_KKPT)	1m 57s
build-android (YANEURAOU_ENGINE_MATERIAL)	1m 46s
build-android (YANEURAOU_MATE_ENGINE)	1m 38s
build-android (TANUKI_MATE_ENGINE)	1m 31s
build-android (USER_ENGINE)	1m 39s
	12m 20s

Provide feedback