Added DataFrame <-> NamedArray type conversion and tests.

[dietercastel]

This includes a dependency on DataFrame of course. Example usage below.

using Test, DataFrames
include("test/init-namedarrays.jl")
julia> n
2×4 Named Array{Float64,2}
A ╲ B │         a          b          c          d
──────┼───────────────────────────────────────────
one   │  0.018457  0.0866435   0.738128  0.0961264
two   │  0.458486   0.629748   0.380246   0.785477

julia> cn = convert(NamedArray,n)
2×4 Named Array{Float64,2}
A ╲ B │         a          b          c          d
──────┼───────────────────────────────────────────
one   │  0.018457  0.0866435   0.738128  0.0961264
two   │  0.458486   0.629748   0.380246   0.785477

julia> ccn = convert(NamedArray,cn)
2×4 Named Array{Float64,2}
A ╲ B │         a          b          c          d
──────┼───────────────────────────────────────────
one   │  0.018457  0.0866435   0.738128  0.0961264
two   │  0.458486   0.629748   0.380246   0.785477

@test n == ccn

[dietercastel]

This PR becomes moot if #99 is merged.

[sylvaticus]

I think you want here :
deleteat!(newdimnames,findfirst(x->x==valueCol,newdimnames))
rather than:
deleteat!(newdimnames,findfirst(x->x==:Values,newdimnames))