Add lots more TensorPrimitives operations

[stephentoub]

All are functional and tested, some are vectorized, others still need to be (subsequent to this PR, tracked by #97193).

Fixes #96451 (still needs final API review)
Fixes #94553 (handles the IndexOfXx methods that were missing)
Fixes #97190
Contributes to #93286

[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.

[ghost]

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

Issue Details

---

All are functional and tested, some are vectorized, others still need to be (subsequent to this PR).

Fixes #96451 (still needs final API review)
Fixes #94553 (handles the IndexOfXx methods that were missing)
Fixes #97190

Author:	stephentoub
Assignees:	-
Labels:	area-System.Numerics, new-api-needs-documentation
Milestone:	-

[stephentoub]

@tannergooding, in addition to normal review stuff, I'd appreciate it if you could triple-check:

• These are the right overloads to be adding.
• The generic constraints on each are correct.
• The method names are what's desired (e.g. I wasn't sure about the MultiplyAddEstimate name, since we call it FusedMultiplyAdd on the interface)

[tannergooding]

nit: These can be ElementwiseSelect since we know sameSign is AllBitsSet or Zero per-element

[stephentoub]

Changed. Separately, we can go through and replace all ConditionalSelects to be ElementWiseSelect as long as the mask is known to be AllBitsSet or Zero, in particular if it's the result of some comparison operation like Equal, LessThan, GreaterThan, etc., right?

[stephentoub]

(Should we expose ElementWiseSelect on the VectorXX types if it's actually beneficial to use?)

[stephentoub]

(And/or make the JIT recognize some cases where it knows the bit pattern is applicable and do the better thing when possible.)

[stephentoub]

Failure in the "runtime (Build browser-wasm linux Release LibraryTests_EAT)" leg is ##[error] ,##[warning]Received request to deprovision: The request was cancelled by the remote provider.

[MichalStrehovsky]

We're seeing failures in main after this merged, see #97087:

Looks like this PR didn't run the legs where this is failing for the Arcade update.

        /_/src/libraries/System.Numerics.Tensors/tests/TensorPrimitives.Generic.cs(215,0): at System.Numerics.Tensors.Tests.GenericFloatingPointNumberTensorPrimitivesTests`1.SpanDestinationFunctions_SpecialValues(SpanDestinationDelegate tensorPrimitivesMethod, Func`2 expectedMethod)
           at InvokeStub_GenericFloatingPointNumberTensorPrimitivesTests`1.SpanDestinationFunctions_SpecialValues(Object, Span`1)
           at System.Reflection.MethodBaseInvoker.InvokeWithFewArgs(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)
    System.Numerics.Tensors.Tests.DoubleGenericTensorPrimitives.SpanDestinationFunctions_ValueRange(tensorPrimitivesMethod: SpanDestinationDelegate { Method = Void ReciprocalEstimate[Double](System.ReadOnlySpan`1[System.Double], System.Span`1[System.Double]), Target = null }, expectedMethod: Func`2 { Method = Double ReciprocalEstimate(Double), Target = null }) [FAIL]
      Assert.All() Failure: 33 out of 201 items in the collection did not pass.
      [27]:  Item:  Tuple (4, -19)
             Error: Assert.Equal() Failure: Values differ
                    Expected: -0.052490234375
                    Actual:   -0.052631578947368418
      [28]:  Item:  Tuple (4, -16)
             Error: Assert.Equal() Failure: Values differ
                    Expected: -0.0623779296875
                    Actual:   -0.0625