map checks dimensions inconsistently

[cstjean]

I believe that the second example should also be a DimensionMismatch, for consistency.

julia> map(<, [1, 2], [2])
ERROR: DimensionMismatch("dimensions must match")
Stacktrace:
 [1] promote_shape at ./indices.jl:154 [inlined]
 [2] axes(::Base.Iterators.Zip2{Array{Int64,1},Array{Int64,1}}) at ./iterators.jl:291
 [3] _array_for at ./array.jl:611 [inlined]
 [4] collect(::Base.Generator{Base.Iterators.Zip2{Array{Int64,1},Array{Int64,1}},getfield(Base, Symbol("##3#4")){typeof(<)}}) at ./array.jl:624
 [5] map(::Function, ::Array{Int64,1}, ::Array{Int64,1}) at ./abstractarray.jl:2060
 [6] top-level scope at none:0

julia> map(<, [1 2], [2])
1-element Array{Bool,1}:
 true

[BioTurboNick]

My issue #33635 is related although different. Scalar inputs are treated like iterables with 1 element, and that issue documents other potentially inconsistent behavior with e.g. generators and tuples as arguments.

As a user, I would wish map()/pmap() to enforce size wherever possible. If a user desires the iterable-as-many-as-possible behavior, I'd argue they should be explicit about that.

[mschauer]

Resolved by #29927