Storage type promotion

[arnavk23]

I fixed the "storage types cannot be promoted to a concrete type" error (issue #329) by making the code robust when promote_type(storage_type(op1), storage_type(op2)) yields a non-concrete type. The fix:

In [operations.jl] : (operator * and +)
In [cat.jl] : (operator hcat and vcat)
Falls back to a concrete operand storage type if promotion yields an abstract type, or uses Vector{T} as a last resort.
This allows GPU-backed operators and mixed-storage scenarios to work without hard failures.

Closes #329

[arnavk23]

@tmigot please review this pr.

[tmigot]

What does the example in the issue do with this PR?

[arnavk23]

@tmigot julia> using CUDA, NLPModels, NLPModelsTest, LinearOperators

julia> V = CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}
CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}

julia> nlp = BROWNDEN(V)
BrowndenNLPModel - Model with 4 variables and 0 constraints

julia> nvar = nlp.meta.nvar
4

julia> xc = V(undef, nvar)
4-element CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}:
0.0
0.0
0.0
0.0

julia> Hs = V(undef, nvar)
4-element CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}:
0.0
0.0
0.0
0.0

julia> H = hess_op!(nlp, xc, Hs)
Linear operator
nrow: 4
ncol: 4
eltype: Float64
symmetric: true
hermitian: true

julia> cg_op_diag = V(undef, nvar)
4-element CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}:
0.0
0.0
0.0
0.0

julia> cg_op = opDiagonal(cg_op_diag)
Linear operator
nrow: 4
ncol: 4
eltype: Float64
symmetric: true
hermitian: false

julia> ZHZ = cg_op' * H * cg_op
Linear operator
nrow: 4
ncol: 4
eltype: Float64
symmetric: false
hermitian: false

julia> size(ZHZ)
(4, 4)

julia> typeof(ZHZ)
LinearOperator{Float64, Int64, var"#..."#..., var"#..."#..., var"#..."#..., CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}}

julia> v = V(undef, nvar); v .= 1.0;

julia> result = ZHZ * v
4-element CuArray{Float64, 1, CUDA.Mem.DeviceBuffer}:
0.0
0.0
0.0
0.0

[codecov]

Codecov Report

❌ Patch coverage is 61.53846% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 93.77%. Comparing base (32dbc5e) to head (3dfa5ff).
⚠️ Report is 48 commits behind head on main.

Files with missing lines	Patch %	Lines
src/abstract.jl	44.44%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #382      +/-   ##
==========================================
- Coverage   95.00%   93.77%   -1.23%     
==========================================
  Files          17       20       +3     
  Lines        1100     1156      +56     
==========================================
+ Hits         1045     1084      +39     
- Misses         55       72      +17     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.