Make AbstractVariable a subtype of DiskArray.

[lupemba]

This is an issue to track my attempt to make abstract type AbstractVariable{T,N} <: AbstractDiskArray{T, N} end motivated by the discussion in issue JuliaGeo/NCDatasets.jl#274

This is not a small task and it will take some time since I am only working on it in my free time.
@tiemvanderdeure have already started some work on the DiskArray side. JuliaIO/DiskArrays.jl#255

[lupemba]

I have made some progress in my first attempt but I am constantly hitting small persistent bugs and corner cases. The latest challenge being JuliaIO/DiskArrays.jl#259

[rafaqz]

What other bugs are there besides that issue?

(which is now fixed here: JuliaIO/DiskArrays.jl#260)

[lupemba]

Hi @rafaqz

Maybe bugs was a bad word to describe it. Most of it is just corner cases that fails in the test.
Example of a corner case I solved yesterday

defVar(ds,"scalar",123.)
@test ds["scalar"][1] == 123.

Example of something I still have to solve is indexing a variable with a CF string. I probably removed a method to much.
height[CF"longitude"][:]

I will create issues if I find any more real bugs.

[Alexander-Barth]

Thanks again @lupemba for your great work! I hope we can celebrate that in the Azores :-)
Can this issue be closed?

[lupemba]

Yes, I hope that they do the JuliaEO again in January, alternatively I will probably stop by JuliaCOn26 since it is close to EUMETSAT :D
The ...Datasets.jl packages still needs to be updated but we can work on that once CommonDataModel v0.4 is released. I close this issue now.