dropdims fails

This would work if A were a normal matrix:

julia> A = @SMatrix randn(2,2)
2×2 SMatrix{2, 2, Float64, 4} with indices SOneTo(2)×SOneTo(2):
 -0.663672  2.31438
  0.166198  1.26303

julia> dropdims(mean(A; dims=2); dims=2)
ERROR: TypeError: in typeassert, expected Tuple{SOneTo{1}}, got a value of type Tuple{SOneTo{2}}
Stacktrace:
 [1] _dropdims(A::SMatrix{2, 1, Float64, 2}, dims::Tuple{Int64})
   @ Base ./abstractarraymath.jl:92
 [2] _dropdims(A::SMatrix{2, 1, Float64, 2}, dim::Int64)
   @ Base ./abstractarraymath.jl:94
 [3] dropdims(A::SMatrix{2, 1, Float64, 2}; dims::Int64)
   @ Base ./abstractarraymath.jl:82
 [4] top-level scope
   @ REPL[26]:1

Reproduce the issue in the Julia REPL using the shown @SMatrix randn(2,2) and dropdims(mean(A; dims=2); dims=2) call. Compare the result with a normal matrix and inspect the StaticArrays handling around dropdims; done means the call no longer raises the shown TypeError and returns the expected reduced array.