Add promote_eltype for CuArray / ROCArray / MtlArray

`ArrayInterface.promote_eltype` only had a method for plain `Array{T, N}`
(with an explicit "no generic fallback is given" note in the docstring).
Downstream packages that pass GPU array types through
`promote_eltype` therefore hit a `MethodError` — for example,
SciML/NonlinearSolve.jl#910 tripped this on
`test/cuda_tests.jl:33 "GeneralizedFirstOrderAlgorithm"` when deriving
a Dual-eltype wrapper-signature array type for `CuArray{Float32}`:

    MethodError: no method matching promote_eltype(
        ::Type{CuArray{Float32, 1, CUDA.DeviceMemory}},
        ::Type{ForwardDiff.Dual{Tag{NonlinearSolveBase.NonlinearSolveTag, Float32}, Float32, 1}})

Adds the obvious eltype-swapping method in each GPU extension,
preserving the non-eltype type parameters (`M` for `CuArray` memory
kind, `B` for `ROCArray` buffer type, `S` for `MtlArray` storage mode):

    ArrayInterface.promote_eltype(
        ::Type{<:CuArray{T, N, M}}, ::Type{T2}
    ) where {T, N, M, T2} = CuArray{promote_type(T, T2), N, M}

    ArrayInterface.promote_eltype(
        ::Type{<:ROCArray{T, N, B}}, ::Type{T2}
    ) where {T, N, B, T2} = ROCArray{promote_type(T, T2), N, B}

    ArrayInterface.promote_eltype(
        ::Type{<:MtlArray{T, N, S}}, ::Type{T2}
    ) where {T, N, S, T2} = MtlArray{promote_type(T, T2), N, S}

Bumps patch version 7.23.0 → 7.24.0 so downstream packages can
compat-bound the new method.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ChrisRackauckas

Project.toml

@@ -1,6 +1,6 @@
 name = "ArrayInterface"
 uuid = "4fba245c-0d91-5ea0-9b3e-6abc04ee57a9"
-version = "7.23.0"
+version = "7.24.0"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"

ext/ArrayInterfaceAMDGPUExt.jl

@@ -12,4 +12,10 @@ end
 
 ArrayInterface.device(::Type{<:AMDGPU.ROCArray}) = ArrayInterface.GPU()
 
+function ArrayInterface.promote_eltype(
+        ::Type{<:AMDGPU.ROCArray{T, N, B}}, ::Type{T2}
+    ) where {T, N, B, T2}
+    return AMDGPU.ROCArray{promote_type(T, T2), N, B}
+end
+
 end # module

ext/ArrayInterfaceCUDAExt.jl

@@ -13,4 +13,10 @@ end
 
 ArrayInterface.device(::Type{<:CUDA.CuArray}) = ArrayInterface.GPU()
 
+function ArrayInterface.promote_eltype(
+        ::Type{<:CUDA.CuArray{T, N, M}}, ::Type{T2}
+    ) where {T, N, M, T2}
+    return CUDA.CuArray{promote_type(T, T2), N, M}
+end
+
 end # module

ext/ArrayInterfaceMetalExt.jl

@@ -12,4 +12,10 @@ end
 
 ArrayInterface.device(::Type{<:Metal.MtlArray}) = ArrayInterface.GPU()
 
+function ArrayInterface.promote_eltype(
+        ::Type{<:Metal.MtlArray{T, N, S}}, ::Type{T2}
+    ) where {T, N, S, T2}
+    return Metal.MtlArray{promote_type(T, T2), N, S}
+end
+
 end # module