Incorrect/unnecessary overloads

[colinxs]

(copied from Slack)

Here's one example:

Statistics.std(X::AbstractVectorOfSimilarArrays{T,M}; corrected::Bool = true) where {T,M} =
     std(flatview(X); corrected = corrected)
which results in the following behavior:
 julia> A=rand(2,3,4);

 julia> B=nestedview(A, Val(2));

 julia> C=[Array(el) for el in B];

 julia> std(B)
 0.32135578668613923

 julia> std(C)
 2×3 Array{Float64,2}:
  0.163392  0.393621  0.343109
  0.290079  0.365516  0.135774

I think the fix for this one is easy: just remove all the overloads for now. If we continue with the plan to merge SpecialArrays/JuliennedArrays here then how dims is handled changes anyways since they both support slicing along arbitrary dimensions and not just the first M.

[oschulz]

just remove all the overloads for now

Please don't, I have packages that critically depend on these things being optimized for VectorOfSimilarVectors (it's doing the right thing there).

[oschulz]

However, I'd be fine with falling back to the default implementation for anything that can't be easily handled by running the stats functions on the flat view.

[colinxs]

I should've clarified, I meant just removing the incorrect overloads for now, rather than patching them, as the proper fix will change things anyways :)

[oschulz]

So, the whole thing was a bit of a mess. I've cleaned this up now, have a look at #12.

I think I need to submit a few PRs to StatsBase, too, at some point, a lot of things that Statistics supports for vectors-of-vectors StatsBase does not (or incorrectly). See comments in new tests.

[oschulz]

Closing this for now - I believe #12 now ensure same semantics as standard nested arrays for statistics functions (except where not defined for standard arrays). We can always revisit this, of course.

[oschulz]

For future reference, this is also related to JuliaStats/StatsBase.jl#518.