sum and mean of skipmissings don't accept the dims kwarg

[oxinabox]

julia> sum(skipmissing([1 2; missing 3]); dims=2)
ERROR: MethodError: no method matching mapfoldl(::typeof(identity), ::typeof(Base.add_sum), ::Base.SkipMissing{Matrix{Union{Missing, Int64}}}; dims=2)
Closest candidates are:
  mapfoldl(::Any, ::Any, ::Any; init) at reduce.jl:162 got unsupported keyword argument "dims"
Stacktrace:
 [1] kwerr(::NamedTuple{(:dims,), Tuple{Int64}}, ::Function, ::Function, ::Function, ::Base.SkipMissing{Matrix{Union{Missing, Int64}}})
   @ Base ./error.jl:157
 [2] mapreduce(f::Function, op::Function, itr::Base.SkipMissing{Matrix{Union{Missing, Int64}}}; kw::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:dims,), Tuple{Int64}}})
   @ Base ./reduce.jl:289
 [3] sum(f::Function, a::Base.SkipMissing{Matrix{Union{Missing, Int64}}}; kw::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:dims,), Tuple{Int64}}})
   @ Base ./reduce.jl:503
 [4] sum(a::Base.SkipMissing{Matrix{Union{Missing, Int64}}}; kw::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:dims,), Tuple{Int64}}})
   @ Base ./reduce.jl:530
 [5] top-level scope
   @ REPL[6]:

Expected output ios [3; 3]

Similar for mean though error is different, it will need the sum case fixed anyway

julia> mean(skipmissing([1 2; missing 3]); dims=2)
ERROR: MethodError: no method matching mean(::Base.SkipMissing{Matrix{Union{Missing, Int64}}}; dims=2)
Closest candidates are:
...
Stacktrace:
[1] top-level scope

Expected output is a matrix [1.5; 3]

Workaround is use eachslice

julia> reshape(mean.(skipmissing.(eachslice([1 2; missing 3], dims=1))), 1, :)
1×2 Matrix{Float64}:
1.5  3.0

Which isn't terrible TBH. especially if you don't care if you get a vector out, then it is just mean.(skipmissing.(eachslice([1 2; missing 3], dims=1)))

There is an old PR relating to this that seems to have gone stale?
#28027

[nalimilan]

If there's general support for #28027 it shouldn't be too hard to rebase (hopefully).