Regression in Base.vect if the elements have offset axes

[jishnub]

This works on 1.7 and 1.8-beta3:

julia> using OffsetArrays

julia> ri = 2:3;

julia> [Base.IdentityUnitRange(ri), OffsetArray(ri, 2)]
2-element Vector{AbstractVector{Int64}}:
 Base.IdentityUnitRange(2:3)
 2:3 with indices 3:4

but on Version 1.9.0-DEV.353 (2022-04-09):

julia> using OffsetArrays

julia> ri = 2:3;

julia> [Base.IdentityUnitRange(ri), OffsetArray(ri, 2)]
ERROR: DimensionMismatch: axes must agree, got (Base.OneTo(2),) and (OffsetArrays.IdOffsetRange(values=3:4, indices=3:4),)
Stacktrace:
 [1] (::Base.var"#checkaxs#141")(axd::Tuple{Base.OneTo{Int64}}, axs::Tuple{OffsetArrays.IdOffsetRange{Int64, Base.OneTo{Int64}}})
   @ Base ./abstractarray.jl:1123
 [2] copyto_axcheck!
   @ ./abstractarray.jl:1125 [inlined]
 [3] Array
   @ ./array.jl:626 [inlined]
 [4] convert
   @ ./array.jl:617 [inlined]
 [5] setindex!(A::Vector{Vector{Int64}}, x::OffsetVector{Int64, UnitRange{Int64}}, i1::Int64)
   @ Base ./array.jl:966
 [6] copyto!(dest::Vector{Vector{Int64}}, src::Tuple{Base.IdentityUnitRange{UnitRange{Int64}}, OffsetVector{Int64, UnitRange{Int64}}})
   @ Base ./abstractarray.jl:909
 [7] vect(::Base.IdentityUnitRange{UnitRange{Int64}}, ::Vararg{Any})
   @ Base ./array.jl:147
 [8] top-level scope
   @ REPL[3]:1

The destination is chosen as Vector{Vector{Int64}}, which doesn't have offset axes.

[vtjnash]

Sounds like a possibly OffsetArrays bug (it needs to define a promotion rule that supersedes that for AbstractArray)