[llvm] [SLP]Reduce number of alternate instruction, where possible (PR #123360)

[Hans Wennborg via llvm-commits]

zmodem wrote:

The re-land is causing assertion failures.

Reproducer: 
[masked_arithmetic_test-b542a4.cpp.gz](https://github.com/user-attachments/files/19207618/masked_arithmetic_test-b542a4.cpp.gz)


```
$ build/bin/clang -cc1 -triple x86_64-unknown-linux-gnu -emit-obj --crel -disable-free -clear-ast-before-backend -main-file-name masked_arithmetic_test.cc -mrelocation-model pic -pic-level 2 -fhalf-no-semantic-interposition -fmerge-all-constants -fno-delete-null-pointer-checks -mframe-pointer=all -relaxed-aliasing -ffp-contract=off -fno-rounding-math -mconstructor-aliases -funwind-tables=2 -target-cpu x86-64 -target-feature +sse3 -tune-cpu generic -debug-info-kind=constructor -dwarf-version=4 -debugger-tuning=gdb -ggnu-pubnames -gsimple-template-names=simple -mllvm -generate-arange-section -debug-forward-template-params -fdebug-compilation-dir=. -ffunction-sections -fdata-sections -fno-unique-section-names -fcoverage-compilation-dir=. -nostdinc++ -DHWY_IS_TEST -D USE_UDEV -D USE_AURA=1 -D USE_GLIB=1 -D USE_OZONE=1 -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D _FORTIFY_SOURCE=2 -D _FILE_OFFSET_BITS=64 -D _LARGEFILE_SOURCE -D _LARGEFILE64_SOURCE -D _GNU_SOURCE -D _LIBCPP_HARDENING_MODE=_LIBCPP_HARDENING_MODE_EXTENSIVE -D COMPONENT_BUILD -D _LIBCPP_INSTRUMENTED_WITH_ASAN=0 -D CR_LIBCXX_REVISION=7c3087b8a7563d8218b27b7183ba41cf52d41508 -D TMP_REBUILD_HACK -D CR_SYSROOT_KEY=20250129T203412Z-1 -D NDEBUG -D NVALGRIND -D DYNAMIC_ANNOTATIONS_ENABLED=0 -D GLIB_VERSION_MAX_ALLOWED=GLIB_VERSION_2_56 -D GLIB_VERSION_MIN_REQUIRED=GLIB_VERSION_2_56 -D UNIT_TEST -D GTEST_API_= -D GTEST_HAS_POSIX_RE=0 -D GTEST_LANG_CXX11=1 -D GTEST_HAS_TR1_TUPLE=0 -D GTEST_HAS_ABSL -D GTEST_NO_ABSL_FLAGS -D ABSL_CONSUME_DLL -D ABSL_FLAGS_STRIP_NAMES=0 -D __DATE__= -D __TIME__= -D __TIMESTAMP__= -D UNSAFE_BUFFERS_BUILD -O2 -std=c++20 -fdeprecated-macro -fvisibility=hidden -fvisibility-inlines-hidden -fwrapv -fwrapv-pointer -pthread -stack-protector 1 -ftrivial-auto-var-init=pattern -fno-rtti -fgnuc-version=4.2.1 -fno-implicit-modules -fskip-odr-check-in-gmf -fno-sized-deallocation -Qn -vectorize-loops -vectorize-slp -fuse-ctor-homing -mllvm -instcombine-lower-dbg-declare=0 -mllvm -split-threshold-for-reg-with-hint=0 -fcomplete-member-pointers -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -x c++ masked_arithmetic_test-b542a4.cpp -w

clang: /work/llvm-project/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp:16237: Value *llvm::slpvectorizer::BoUpSLP::vectorizeTree(TreeEntry *): Assertion `OpTE1.isSame( ArrayRef(E->Scalars).take_front(OpTE1.getVectorFactor())) && "Expected same first part of scalars."' failed.
```

I'll revert back to green.

https://github.com/llvm/llvm-project/pull/123360