Don't throw from TensorPrimitives.Clamp when min is greater than max

[tannergooding]

The span-based TensorPrimitives.Clamp overloads dispatch to the ClampOperator* structs whose scalar Invoke calls T.Clamp, which throws ArgumentException when min > max, while the Vector128/256/512 Invoke paths call Vector*.Clamp, which follow HLSL and compute Min(Max(x, min), max) without validation.

This made behavior position-dependent for the span-min/span-max overloads: an unordered min[i] > max[i] pair threw when it landed in the scalar remainder but was silently clamped when it landed in the vectorized region. The Clamp(ReadOnlySpan<T> x, T min, T max, ...) overload additionally validated min <= max up front and threw before dispatch.

---

Align the scalar path with the vector path by computing Min(Max(x, min), max) in the scalar Invoke and removing the up-front validation, so Clamp succeeds for unordered bounds regardless of element position, matching Vector.Clamp. T.Min(T.Max(x, min), max) is exactly T.Clamp minus the throw, and (via IEEE Min/Max) agrees with the vector path on NaN and signed zeros by construction.

Also removes the now-unused ThrowArgument_MinGreaterThanMax helper and its resource string, and updates the doc comments.

Tests: adds Clamp_UnorderedBoundsMatchScalarSemantics and Clamp_SpecialValuesMatchScalarSemantics to GenericNumberTensorPrimitivesTests<T>, exercising both the scalar (sub-vector lengths) and vectorized (length 128) paths across all affected overloads/operand orderings and types, including NaN and signed-zero parity between the two paths for the floating-point types.

---

Note

This is a documented-behavior change: the previously documented ArgumentException on min > max is removed, so it warrants breaking-change consideration.

Note

This PR was authored with GitHub Copilot.

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).
12 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

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

[Copilot]

Pull request overview

This PR changes TensorPrimitives.Clamp to avoid throwing when min > max by aligning the scalar fallback behavior with the vectorized Vector128/256/512.Clamp semantics (compute Min(Max(x, min), max)), and updates documentation/tests accordingly.

Changes:

• Remove the scalar min > max validation and switch scalar ClampOperator* implementations to T.Min(T.Max(x, min), max).
• Delete the now-unused throw helper and resource string for the removed ArgumentException.
• Add generic tests to ensure scalar vs vectorized paths match (including floating-point special values like NaN and signed zero).

Show a summary per file

File	Description
src/libraries/System.Numerics.Tensors/src/System/Numerics/Tensors/netcore/TensorPrimitives.Clamp.cs	Removes min > max throwing and makes scalar operator semantics match vector clamp; updates XML docs.
src/libraries/System.Numerics.Tensors/src/System/ThrowHelper.cs	Removes unused ThrowArgument_MinGreaterThanMax.
src/libraries/System.Numerics.Tensors/src/Resources/Strings.resx	Removes unused Argument_MinGreaterThanMax resource string.
src/libraries/System.Numerics.Tensors/tests/TensorPrimitives.Generic.cs	Adds regression tests validating consistent clamp semantics across scalar/vectorized paths and special values.

Copilot's findings

• Files reviewed: 4/4 changed files
• Comments generated: 1

[Copilot]

Copilot's findings

• Files reviewed: 4/4 changed files
• Comments generated: 0 new

[Copilot]

Copilot's findings

• Files reviewed: 4/4 changed files
• Comments generated: 4

[Copilot]

Copilot's findings

• Files reviewed: 4/4 changed files
• Comments generated: 3

[Copilot]

Copilot's findings

• Files reviewed: 4/4 changed files
• Comments generated: 0 new