Investigate microbenchmarks that regress with PGO enabled #84264

Look into cases where microbenchmarks are slower with PGO enabled.

#74873 (comment) has an initial list to check.

Issues uncovered

• Microbenchmarks where inliner runs out of budget with TieredPGO #85531
• PGO: redundant Tier0-instr promotions #84517
• Suspect -p ETW is not enabling the CLR rundown provider BenchmarkDotNet#2296
• Methods with both stackalloc and loops bypass tiering #85548
• JIT: widening stack memory loads can cause store-forward stalls #85957

Fixes

• JIT: use synthesis to repair some reconstruction issues #84312
• Fix simd12_t size #85362
• JIT: extend initial profile repair to more cases #84741
• Keep instrumentings some intrinsics #85130
• JIT: use blend rather then repair for profile inconsistencies #85171
• JIT: improve profile update for loop inversion #85265
• JIT: prepare for instrumentation before incorporating profile counts #85805
• JIT: produce consistent flow graph when producing or consuming profile data #85860
• JIT: fix another case of early flow graph divergence #85873
• JIT: assert if we see schema mismatches with dynamic pgo data #85898