Investigate failure on latest ODE deps

[visr]

#2070 needed a2d39f4 before it could be merged. Whilst the error message involves a MethodError for reshape of a ComponentVector, it doesn't seem related to recent ComponentArray changes, as found in 211501b. Moving back from LinearSolve 3 to 2 brings many packages back, but allowed us to merge #2070. We can try again later and if it still happens try to find the reason and report upstream.

MethodError: no method matching  reshape(::Vector{Int64}, ::Tuple{ComponentArrays.CombinedAxis{ComponentArrays.Axis{(tabulated_rating_curve = 1:3, pump = 4:4, outlet = 5:4, user_demand_inflow = 5:4, user_demand_outflow = 5:4, linear_resistance = 5:6, manning_resistance = 7:7, evaporation = 8:11, infiltration = 12:15, integral = 16:15)}, Base.OneTo{Int64}}})\nThe function `reshape` exists, but no method is defined for this combination of argument types

[visr]

Here is a stacktrace. It does seem ComponentArrays related, so I expect #2077 to help this as well.

This reshape fails because ComponentArrays returns custom axes(x) which don't have a method it seems.

https://github.com/JuliaDiff/FiniteDiff.jl/blob/a7eca2d4b73c4de12140d89df7621fcc90d29190/src/jacobians.jl#L422

ERROR: MethodError: no method matching reshape(::Vector{Int64}, ::Tuple{ComponentArrays.CombinedAxis{ComponentArrays.Axis{…}, Base.OneTo{…}}})
The function `reshape` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  reshape(::Array{T, M}, ::NTuple{N, Int64}) where {T, N, M}
   @ Base reshapedarray.jl:40
  reshape(::AbstractArray, ::Tuple{Vararg{Union{Colon, Int64}}})
   @ Base reshapedarray.jl:129
  reshape(::AbstractArray, ::NTuple{N, Int64} where N)
   @ Base reshapedarray.jl:122
  ...

Stacktrace:
  [1] reshape(parent::Vector{Int64}, dims::ComponentArrays.CombinedAxis{ComponentArrays.Axis{…}, Base.OneTo{…}})
    @ Base .\reshapedarray.jl:120
  [2] finite_difference_jacobian!(J::SparseArrays.SparseMatrixCSC{…}, f::SciMLBase.UJacobianWrapper{…}, x::ComponentArrays.ComponentVector{…}, cache::FiniteDiff.JacobianCache{…}, f_in::ComponentArrays.ComponentVector{…}; relstep::Float64, absstep::Float64, colorvec::Vector{…}, sparsity::SparseArrays.SparseMatrixCSC{…}, dir::Int64)
    @ FiniteDiff a:\.julia\packages\FiniteDiff\EBPBu\src\jacobians.jl:419
  [3] jacobian_finitediff_forward!(J::SparseArrays.SparseMatrixCSC{…}, f::Function, x::ComponentArrays.ComponentVector{…}, jac_config::FiniteDiff.JacobianCache{…}, forwardcache::ComponentArrays.ComponentVector{…}, integrator::OrdinaryDiffEqCore.ODEIntegrator{…})
    @ OrdinaryDiffEqDifferentiation a:\.julia\packages\OrdinaryDiffEqDifferentiation\6Bzim\src\derivative_wrappers.jl:219

[visr]

This MethodError is fixed by SciML/ComponentArrays.jl#295. I made #2083 to check if that is enough to update the other packages.