JIT: examine need for canInlineTypeCheckWithObjectVTable

[AndyAyersMS]

When the jit is optimizing type equality and inequality tests, it may run into the following case:

    object.GetType() == typeof(X)

This can usually be optimized into a comparison of the object's method table pointer versus a known method table address. But historically that has not always been a valid optimization. So the jit calls canInlineTypeCheckWithObjectVTable to identify the cases where the optimization is valid.

https://github.com/dotnet/coreclr/blob/3c01252879f2d2223f63d5b222b6584c28fc07f4/src/jit/gentree.cpp#L13141-L13147

Per Jan's comments in #10332, it seems in CoreCLR this optimization is likely always valid and the check is no longer needed.

category:cq
theme:basic-cq
skill-level:intermediate
cost:small

[MichalStrehovsky]

It might be nice to keep the check, but have this always return true on CoreCLR.

CoreRT has things like "frozen objects" (where an object instance can be preallocated in the PE image instead of on the GC heap - so e.g. we never have to allocate string literals on the GC heap) - for those, the method table pointer can point to things that can't be compared using simple method table equality (method tables are not equal, but they are the same object type).

[AndyAyersMS]

Makes sense.

[AndyAyersMS]

This interface method is no longer needed, see #2276 and #32521.