Expose the Sin, Cos, and SinCos methods on the Vector types

[tannergooding]

This makes additional progress towards #93513

This implementation is more accurate than the naive implementation that TensorPrimitives had been using and correctly handles the full range of inputs. It does still fallback to scalar if any input is greater than 5 million as vectorizing the complex reduction is more involved and I want to handle it separately in .NET 10.

[dotnet-issue-labeler]

Note regarding the new-api-needs-documentation label:

This serves as a reminder for when your PR is modifying a ref *.cs file and adding/modifying public APIs, please make sure the API implementation in the src *.cs file is documented with triple slash comments, so the PR reviewers can sign off that change.

[dotnet-issue-labeler]

Note regarding the new-api-needs-documentation label:

This serves as a reminder for when your PR is modifying a ref *.cs file and adding/modifying public APIs, please make sure the API implementation in the src *.cs file is documented with triple slash comments, so the PR reviewers can sign off that change.

[dotnet-policy-service]

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

[michaelgsharp]

LGTM

[tannergooding]

This is blocked on the JIT side fix going in (#104935)

[tannergooding]

Noting that the regression is "by design" in that we took an accuracy fix to ensure we were overall consistent with the scalar algorithm

If the input is less then 5 million then we are generally fast and if its less PI / 4 (already reduced) then we are very fast. However, if the input is greater than 5 million we fallback to scalar currently. This is something we can fix, but it was out of scope for .NET 10 since such inputs should be rare.