Use AxisArrays for distance/proximity matrices

[andreasnoack]

That would make it easier to maintain the association between the original grid and the distance values. E.g. it would allow for something like

julia> g = ConScape.Grid(size(r)..., affinities=a, costs=_cost, prune=true)
[ Info: cost graph contains 6 strongly connected subgraphs
[ Info: removing 8 nodes from affinity and cost graphs
ConScape.Grid of size 4x4

julia> lc = ConScape.least_cost_distance(g)
8×8 Matrix{Float64}:
 0.0       0.693147  1.38629   2.07944   0.693147  1.38629   2.07944   2.77259
 0.693147  0.0       0.693147  1.38629   1.38629   0.693147  1.38629   2.07944
 1.38629   0.693147  0.0       0.693147  2.07944   1.38629   0.693147  1.38629
 2.07944   1.38629   0.693147  0.0       2.77259   2.07944   1.38629   0.693147
 1.38629   2.07944   2.77259   3.46574   0.0       1.38629   2.77259   4.15888
 2.07944   1.38629   2.07944   2.77259   1.38629   0.0       1.38629   2.77259
 2.77259   2.07944   1.38629   2.07944   2.77259   1.38629   0.0       1.38629
 3.46574   2.77259   2.07944   1.38629   4.15888   2.77259   1.38629   0.0

julia> lc_aa = AxisArray(lc, source=vec(CartesianIndices((4,4))[:,3:4]), target=vec(CartesianIndices((4,4))[:,3:4]))
2-dimensional AxisArray{Float64,2,...} with axes:
    :source, CartesianIndex{2}[CartesianIndex(1, 3), CartesianIndex(2, 3), CartesianIndex(3, 3), CartesianIndex(4, 3), CartesianIndex(1, 4), CartesianIndex(2, 4), CartesianIndex(3, 4), CartesianIndex(4, 4)]
    :target, CartesianIndex{2}[CartesianIndex(1, 3), CartesianIndex(2, 3), CartesianIndex(3, 3), CartesianIndex(4, 3), CartesianIndex(1, 4), CartesianIndex(2, 4), CartesianIndex(3, 4), CartesianIndex(4, 4)]
And data, a 8×8 Matrix{Float64}:
 0.0       0.693147  1.38629   2.07944   0.693147  1.38629   2.07944   2.77259
 0.693147  0.0       0.693147  1.38629   1.38629   0.693147  1.38629   2.07944
 1.38629   0.693147  0.0       0.693147  2.07944   1.38629   0.693147  1.38629
 2.07944   1.38629   0.693147  0.0       2.77259   2.07944   1.38629   0.693147
 1.38629   2.07944   2.77259   3.46574   0.0       1.38629   2.77259   4.15888

julia> lc_aa[:, atvalue(CartesianIndex(3,4))]
1-dimensional AxisArray{Float64,1,...} with axes:
    :source, CartesianIndex{2}[CartesianIndex(1, 3), CartesianIndex(2, 3), CartesianIndex(3, 3), CartesianIndex(4, 3), CartesianIndex(1, 4), CartesianIndex(2, 4), CartesianIndex(3, 4), CartesianIndex(4, 4)]
And data, a 8-element Vector{Float64}:
 2.0794415416798357
 1.3862943611198906
 0.6931471805599453
 1.3862943611198906
 2.772588722239781
 1.3862943611198906
 0.0
 1.3862943611198906

[rafaqz]

On dev these are now returned as Rasters.jl Raster (A DimensionalData.jl AbstractDimArray) which means they also track crs and other spatial metadata, and can be e.g. written to a .tiff file directly.