[llvm] [RISCV] Adopt SpacemitX60's scheduling model for `-mtune=generic` (PR #167008)

[Luke Lau via llvm-commits]

lukel97 wrote:

Sorry for the delay, here's the LNT results of this patch on the P550 compiled with -march=rv64gc_zba_zbb -O3 -flto: https://lnt.lukelau.me/db_default/v4/nts/1242?compare_to=1245

On SPEC CPU 2017 525.x264_r is 16% faster and 508.namd_r is 7.6% faster, and there were no regressions. 

There were some regressions in the loop vectorizer microbenchmarks but I don't think they're important given that they're geared towards a target with vector support. 

I'll get the numbers on the Banana Pi F3 itself next.

https://github.com/llvm/llvm-project/pull/167008