Add conversion from AbstractArray to mxarray

[tqml]

Adds a conversion method to convert Julia abstract array to mxarray to avoid having them treated as structs in MATLAB.

Example:

a = transpose(rand(10)) # transpose
x = mxarray(a) # x is an actual matrix in matlab, was a struct before
y = jvalue(x) # the return value is now a matrix instead of a dictionary

Related issue #196

[tqml]

I removed the definitions for Ranges and BitArrays since they are only be a special case of AbstractArray, would unnecessary allocate data and get copied to MATLAB anyway

Old defs:

mxarray(a::BitArray) = mxarray(convert(Array{Bool}, a))
mxarray(a::AbstractRange) = mxarray([a;])

New definition was tested with additional special matrix types such as Diagonal, transpose and LowerTriangular.

function mxarray(a::AbstractArray{T}) where {T<:MxRealNum}
    mx = mxarray(T, size(a))
    ptr = data_ptr(mx)
    na = length(a)
    dat = unsafe_wrap(Array{T}, ptr, na)
    for (i, ix) in enumerate(eachindex(a))
        dat[i] = a[ix]
    end
    return mx
end