Add support for storing method handle histograms in profiles

[jakobbotsch]

Allow method handle histograms in .mibc files and in the PGO text format.

Contributes to #44610.

@davidwrighton would this in theory need a R2R version bump? I suppose we can wait with such a bump until we actually start producing profiles with this data (at which point we should also add an equivalent of GetLikelyClass)?

cc @dotnet/jit-contrib

[AndyAyersMS]

Jit and SPMI changes look good.

WIll defer to David or Michal for the remainder of the changes.

[BruceForstall]

Do you need to change the JIT-EE GUID?

[jakobbotsch]

Do you need to change the JIT-EE GUID?

I have some follow-up changes that will actually start producing and using the new kind of PGO data. I think we can delay updating the JIT-EE GUID to that as just adding new PGO schema kinds should be binary compatible (and my future changes will probably also need another JIT-EE GUID update).

[davidwrighton]

@jakobbotsch Yes, this requires a bump in the R2R minor version.

[davidwrighton]

@jakobbotsch Have you verified that with these changes the dotnet-pgo tool continues to work with .NET 6 runtimes?

[jakobbotsch]

@jakobbotsch Yes, this requires a bump in the R2R minor version.

Is it ok to wait until future changes that actually start producing and consuming these? At that point I will add something similar to GetLikelyClass for methods which will require another R2R version bump.

@jakobbotsch Have you verified that with these changes the dotnet-pgo tool continues to work with .NET 6 runtimes?

Yes, it produced a .mibc with the same hash as dotnet-pgo from current main in my test.

[davidwrighton]

When in doubt, bump the version. If there is no chance that new data could be produced, you don't need to, but bumping the R2R version is pretty cheap, and doesn't come with significant costs as old R2R code should not be broken.

[jakobbotsch]

@davidwrighton I've bumped the R2R minor version and also the JIT-EE GUID to be safe, please take another look.
I plan to wait with merging this until after the snap in any case.

[jakobbotsch]

ping @davidwrighton, anything else you think I should do here?

[AndyAyersMS]

Just looked at JIT / SPMI in depth.

[jakobbotsch]

Ping @davidwrighton, my follow-up depends on this one.