Tensor.StdDev does not handle non-contiguous input

[lilinus]

Description

Tensor.StdDev does not handle non-contiguous input.

The method uses

Reproduction Steps

using System.Numerics.Tensors;

// Create 4x4 matrix with upper-left 2x2 matrix
var t = Tensor.Create<double>([4, 4]);
t[[0, 0]] = 1;
t[[0, 1]] = 1;
t[[1, 0]] = 1;
t[[1, 1]] = 1;

// Take slice of upper left 2x2 matrix
var t1 = t.Slice([0..2, 0..2]); // This makes a copy so it is contiguous
var t2 = t.AsReadOnlyTensorSpan().Slice([0..2, 0..2]); // Refers to original span, i.e. non-cotiguous.

// Should print 0 (StdDev of all ones)
Console.WriteLine(Tensor.StdDev<double>(t1));
Console.WriteLine(Tensor.StdDev<double>(t2));

Expected behavior

Both print statements should print 0.

Actual behavior

First print statement prints 0.
Second print throws: System.ArgumentException: 'Destination is too short. Arg_ParamName_Name'

Regression?

No response

Known Workarounds

No response

Configuration

.NET 9

Other information

No response

[dotnet-policy-service]

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

[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.

[jeffhandley]

Assigned to @tannergooding for triage.