Fix CI docs build

[abhro]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the contributor guidelines, in particular the SciML Style Guide and COLPRAC.
• Any new documentation only uses public API

Additional context

Fixes errors in buildkite CI docs build.

Currently in draft mode: the example in tutorials/bvp_example.md (Ascher method for solving BVDAEs) fails on the newest LinearSolve.jl (3.47.0), but if pinned to LinearSolve.jl v3.40.0, it does produce results. Not sure what the underlying issues are yet.

[abhro]

The BVP Ascher thing fails on LinearSolve v3.46.0, but works with 3.45.0. MWE script:

using Pkg
Pkg.activate(temp=true)
ENV["JULIA_PKG_PRECOMPILE_AUTO"] = false
Pkg.add(name="LinearSolve", version=ARGS[1])
Pkg.add("BoundaryValueDiffEq")

import BoundaryValueDiffEq as BVP
Pkg.status()
function f!(du, u, p, t)
    e = 2.7
    du[1] = (1 + u[2] - sin(t)) * u[4] + cos(t)
    du[2] = cos(t)
    du[3] = u[4]
    du[4] = (u[1] - sin(t)) * (u[4] - e^t)
end
function bc!(res, u, p, t)
    res[1] = u[1]
    res[2] = u[3] - 1
    res[3] = u[2] - sin(1.0)
end
u0 = [0.0, 0.0, 0.0, 0.0]
tspan = (0.0, 1.0)
fun = BVP.BVPFunction(f!, bc!, mass_matrix = [1 0 0 0; 0 1 0 0; 0 0 1 0; 0 0 0 0])
prob = BVP.BVProblem(fun, u0, tspan)
sol = BVP.solve(prob,
    BVP.Ascher4(; zeta = [0.0, 0.0, 1.0], jac_alg = BVP.BVPJacobianAlgorithm(BVP.AutoForwardDiff()));
    dt = 0.01)

run as: julia script.jl X.Y.Z where X.Y.Z is the LinearSolve.jl version to use

[abhro]

The problems from DiffEqProblemLibrary.jl appear to be the older ones. I think a new version is needed for (ODEProblemLibrary, SDEProblemLibrary, DDEProblemLibrary, DAEProblemLibrary, BVProblemLibrary)

[ChrisRackauckas]

What's the error? We should make the docs build fail on error

[abhro]

Running with 3.46 gives

$ julia mwe.jl 3.46.0
  Activating new project at `/tmp/jl_eW8Fpj`
   Resolving package versions...
   Installed LinearSolve ─ v3.46.0
    Updating `/tmp/jl_eW8Fpj/Project.toml`
⌃ [7ed4a6bd] + LinearSolve v3.46.0
    Updating `/tmp/jl_eW8Fpj/Manifest.toml`
[... Pkg output ...]
   Resolving package versions...
    Updating `/tmp/jl_eW8Fpj/Project.toml`
  [764a87c0] + BoundaryValueDiffEq v5.18.0
    Updating `/tmp/jl_eW8Fpj/Manifest.toml`
[... Pkg output ...]
Precompiling BoundaryValueDiffEq finished.
  15 dependencies successfully precompiled in 425 seconds. 198 already precompiled.
Status `/tmp/jl_eW8Fpj/Project.toml`
  [764a87c0] BoundaryValueDiffEq v5.18.0
⌃ [7ed4a6bd] LinearSolve v3.46.0
Info Packages marked with ⌃ have new versions available and may be upgradable.
┌ Error: BLAS/LAPACK dgetrf: Matrix is singular
│   Details: U(1,1) is exactly zero. The factorization has been completed, but U is singular and division by U will produce infinity.
│   Return code (info): 1
│   matrix_size: (303, 303)
│   rhs_type: Vector{Float64}
│   rhs_size: (303,)
│   memory_usage_MB: 0.7
│   matrix_type: Matrix{Float64}
│   element_type: Float64
└ @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:267
ERROR: LoadError: BLAS/LAPACK dgetrf: Matrix is singular
  Details: U(1,1) is exactly zero. The factorization has been completed, but U is singular and division by U will produce infinity.
  Return code (info): 1
  matrix_size: (303, 303)
  rhs_type: Vector{Float64}
  rhs_size: (303,)
  memory_usage_MB: 0.7
  matrix_type: Matrix{Float64}
  element_type: Float64
Stacktrace:
  [1] emit_message(message::String, level::Base.CoreLogging.LogLevel, file::String, line::Int64, _module::Module)
    @ SciMLLogging ~/.julia/packages/SciMLLogging/3uoc4/src/utils.jl:57
  [2] macro expansion
    @ ~/.julia/packages/SciMLLogging/3uoc4/src/utils.jl:147 [inlined]
  [3] solve!(cache::LinearSolve.LinearCache{Matrix{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, LinearSolve.DefaultLinearSolver, [types elided]...)
    @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:267
  [4] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:239 [inlined]
  [5] macro expansion
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:514 [inlined]
  [6] solve!(::LinearSolve.LinearCache{Matrix{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, LinearSolve.DefaultLinearSolver, [types elided]...)
    @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:503
  [7] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:503 [inlined]
  [8] #solve!#21
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/common.jl:441 [inlined]
  [9] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/common.jl:440 [inlined]
 [10] #_#1
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/ext/NonlinearSolveBaseLinearSolveExt.jl:24 [inlined]
 [11] LinearSolveJLCache
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/ext/NonlinearSolveBaseLinearSolveExt.jl:14 [inlined]
 [12] solve!(cache::NonlinearSolveBase.NewtonDescentCache{Vector{Float64}, Nothing, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/descent/newton.jl:119
 [13] solve! (repeats 2 times)
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/descent/newton.jl:94 [inlined]
 [14] step!(cache::NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithmCache{Vector{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveFirstOrder ~/.julia/packages/NonlinearSolveFirstOrder/inFag/src/solve.jl:266
 [15] step!
    @ ~/.julia/packages/NonlinearSolveFirstOrder/inFag/src/solve.jl:244 [inlined]
 [16] #step!#180
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:559 [inlined]
 [17] step!
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:553 [inlined]
 [18] solve!(cache::NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithmCache{Vector{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:287
 [19] #__solve#155
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:273 [inlined]
 [20] __solve
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:270 [inlined]
 [21] macro expansion
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:478 [inlined]
 [22] __generated_polysolve(::SciMLBase.NonlinearProblem{Vector{Float64}, true, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:423
 [23] __generated_polysolve
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:423 [inlined]
 [24] #__solve#166
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:396 [inlined]
 [25] __perform_ascher_iteration(cache::BoundaryValueDiffEqAscher.AscherCache{true, Float64, SciMLBase.BVProblem{Vector{Float64}, Tuple{Float64, Float64}, [types elided]...)
    @ BoundaryValueDiffEqAscher ~/.julia/packages/BoundaryValueDiffEqAscher/CPeel/src/ascher.jl:162
 [27] #__solve#34
    @ ~/.julia/packages/BoundaryValueDiffEqCore/hFduL/src/BoundaryValueDiffEqCore.jl:38 [inlined]
 [28] __solve
    @ ~/.julia/packages/BoundaryValueDiffEqCore/hFduL/src/BoundaryValueDiffEqCore.jl:35 [inlined]
 [29] #solve_call#23
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:127 [inlined]
 [30] solve_call
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:84 [inlined]
 [31] solve_up(prob::SciMLBase.BVProblem{Vector{Float64}, Tuple{Float64, Float64}, true, Nothing, SciMLBase.NullParameters, SciMLBase.BVPFunction{true, [types elided]...)
    @ DiffEqBase ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:563
 [32] solve_up
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:540 [inlined]
 [33] #solve#29
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:530 [inlined]
 [34] top-level scope
    @ ~/mwe.jl:25
 [35] include(mod::Module, _path::String)
    @ Base ./Base.jl:306
 [36] exec_options(opts::Base.JLOptions)
    @ Base ./client.jl:317
 [37] _start()
    @ Base ./client.jl:550
in expression starting at mwe.jl:25