Enable vectorization of TensorPrimitives.ConvertTruncating for float/double

[stephentoub]

These were disabled against a PR that has already been merged.

[Copilot]

Pull Request Overview

This PR re-enables vectorization for the ConvertTruncating implementations of float→int, float→uint, double→ulong, and double→long by changing their Vectorizable flags from false to true.

• Flip Vectorizable from false to true for four conversion operators
• Removes outdated TODO comments
• Allows JIT to use SIMD intrinsics for these conversions

Comments suppressed due to low confidence (1)

src/libraries/System.Numerics.Tensors/src/System/Numerics/Tensors/netcore/TensorPrimitives.ConvertTruncating.cs:106

• Since vectorization is now enabled, add unit tests for the float→int, float→uint, double→ulong, and double→long truncating paths (including edge cases like NaN, ±∞, and out-of-range values) to ensure the SIMD implementation matches the scalar behavior.

            public static bool Vectorizable => true;

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-numerics-tensors
See info in area-owners.md if you want to be subscribed.