Improve documentation for CPU target and feature selection

[ScottTodd]

There has been a lot of great work recently on CPU configuration across multiple CPU architectures and specific device models. We should distill what we've collectively learned into the documentation.

This section: https://iree.dev/guides/deployment-configurations/cpu/#compile-and-run-a-program was written some time ago and pretty much says "--iree-llvmcpu-target-triple and --iree-llvmcpu-target-cpu-features are important but we don't fully understand them yet". I think we're starting to understand them :)

Possible topics to cover:

• How to determine what CPU, CPU features, etc. a particular device has
• How to communicate those features to the compiler (flag recipes, links to benchmark configurations, etc.)
• How to verify that generated code made use of certain features
• What to expect out of the box, and why automatic detection is not sufficient for ahead of time scenarios
  • Aside: for JIT scenarios (i.e. TorchDynamo), are we falling short and can we do better?

[ScottTodd]

cc @dcaballe @bjacob (others?)

[ScottTodd]

Doc/tooling updates here would still be helpful

Recent discussion: https://discord.com/channels/689900678990135345/689932666505920596/1225523467215044742

Does anyone know how to check what --iree-llvmcpu-target-cpu-features=host resolves to on a given machine? Trying to match what a user observed on their machine from mine

someone says llc -march=arm -mattr=help
for us, if you pass that flag and compile-to=input it should show you all them in the device config
#executable_target_embedded_elf_x86_64_ = #hal.executable.target<"llvm-cpu", "embedded-elf-x86_64", {cpu = "znver2", cpu_features = "+prfchw,-cldemote,+avx,+aes,+sahf,+pclmul,-xop,+crc32,+xsaves,-avx512fp16,-usermsr,-sm4,-egpr,+sse4.1,-avx512ifma,+xsave,-avx512pf,+sse4.2,-tsxldtrk,-ptwrite,-widekl,-sm3,-invpcid,+64bit,+xsavec,-avx10.1-512,-avx512vpopcntdq,+cmov,-avx512vp2intersect,-avx512cd,+movbe,-avxvnniint8,-avx512er,-ccmp,-amx-int8,-kl,-avx10.1-256,-sha512,-avxvnni,-rtm,+adx,+avx2,-hreset,-movdiri,-serialize,-vpclmulqdq,-avx512vl,-uintr,-cf,+clflushopt,-raoint,-cmpccxadd,+bmi,-amx-tile,+sse,-gfni,-avxvnniint16,-amx-fp16,-ndd,+xsaveopt,+rdrnd,-avx512f,-amx-bf16,-avx512bf16,-avx512vnni,-push2pop2,+cx8,-avx512bw,+sse3,-pku,+fsgsbase,+clzero,+mwaitx,-lwp,+lzcnt,+sha,-movdir64b,-ppx,+wbnoinvd,-enqcmd,-prefetchwt1,-avxneconvert,-tbm,-pconfig,-amx-complex,+ssse3,+cx16,+bmi2,+fma,+popcnt,-avxifma,+f16c,-avx512bitalg,+rdpru,+clwb,+mmx,+sse2,+rdseed,-avx512vbmi2,-prefetchi,+rdpid,-fma4,-avx512vbmi,-shstk,-vaes,-waitpkg,-sgx,+fxsr,-avx512dq,+sse4a", data_layout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128", native_vector_size = 32 : i64, target_triple = "x86_64-unknown-unknown-eabi-elf"}>