Mono incorrectly handles floating-point conversions on 64-bit

[tannergooding]

#69756 has several failures that occur across Mono:

Assert.Throws<OverflowException>(() => NumberBaseHelper<long>.CreateChecked<float>(+9223372036854775808.0f));

Assert.Throws<OverflowException>(() => NumberBaseHelper<ulong>.CreateChecked<double>(+18446744073709551616.0));

In both cases the double and float used is the smallest representable double that is greater than long.MaxValue or ulong.MaxValue, respectively, and therefore it should always overflow.

However, Mono doesn't throw System.OverflowException instead. The same tests exist for x86/x64 where things are passing as expected.

---

#70087 likewise shows issues with conversion of float/double to nuint (System.UIntPtr) where it appears to always be treating the target type as 32-bits, even on 64-bit systems.

I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label.

[tannergooding]

Noting that this was update to include another bug related to double/float to UIntPtr conversions where Mono is treating the target type as 32-bits and which impacts all 64-bit systems, not just Arm64.

[SamMonoRT]

@BrzVlad please take a look
cc @fanyang-mono fyi