JIT: Change IntegralRange to handle arbitrary int64_t

[BoyBaykiller]

Comment on IntegralRange says:

runtime/src/coreclr/jit/compiler.h

Lines 1189 to 1191 in b745957

	// Represents an integral range useful for reasoning about integral casts.
	// It uses a symbolic representation for lower and upper bounds so
	// that it can efficiently handle integers of all sizes on all hosts.

Yet when I change it to use plain int64_t the tp diffs are negligible.

This change will allows us to tighten ranges and potentially enable some optimizations.
The very reason I did this is to be able to handle #126529 (comment)
And just overall its the more natural thing.

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

[EgorBo]

@BoyBaykiller can you fix the conflict?

This change will allows us to tighten ranges and potentially enable some optimizations.

TBH, I'd prefer to slowly move away from IntegralRange and do everything on the VN level

[BoyBaykiller]

TBH, I'd prefer to slowly move away from IntegralRange and do everything on the VN level

Do you mean using existing GetRangeFromAssertions or similar to get range info? Cause VNs alone don't get us range info?
But ye, I also support moving away from IntegralRange 👍