Cube of dataset with a 0-dimensional subcube gives confusing error message

I was trying to write up an example for the JuliaClimate meta issue about julia packages for the CMIP6 data
https://github.com/JuliaClimate/meta/issues/17 and I ran into the following error with trying to convert a CMIP6 dataset into a YAXArray Cube.
The problem is, that the dataset I randomly selected seem to have a zero dimensional subcube which can't be used to derive the size of the expected data cube from the concatenation.

julia> using AWS, Zarr, YAXArrays, CSV, DataFrames

julia> AWS.global_aws_config(AWSConfig(creds=nothing, region=""))
AWSConfig(nothing, "", "json")

julia> # table with all the URLs to the data on google cloud
       CMIP6_stores = CSV.read(download("https://storage.googleapis.com/cmip6/cmip6-zarr-consolidated-stores.csv"), DataFrame)

julia> store = filter(CMIP6_stores) do row
         row.activity_id == "ScenarioMIP" && row.institution_id=="DKRZ" && row.variable_id=="tas" && row.experiment_id=="ssp585"
       end
3×11 DataFrame
 Row │ activity_id  institution_id  source_id      experiment_id  member_id  table_id  variable_id  grid_label  zstore                             dcpp_init_year  version  
     │ String15     String31        String31       String31       String15   String15  String31     String7     String                             Float64?        Int64    
─────┼──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
   1 │ ScenarioMIP  DKRZ            MPI-ESM1-2-HR  ssp585         r1i1p1f1   day       tas          gn          gs://cmip6/CMIP6/ScenarioMIP/DKR…         missing  20190710
   2 │ ScenarioMIP  DKRZ            MPI-ESM1-2-HR  ssp585         r1i1p1f1   Amon      tas          gn          gs://cmip6/CMIP6/ScenarioMIP/DKR…         missing  20190710
   3 │ ScenarioMIP  DKRZ            MPI-ESM1-2-HR  ssp585         r1i1p1f1   3hr       tas          gn          gs://cmip6/CMIP6/ScenarioMIP/DKR…         missing  20190710

julia> zopen(store.

activity_id     dcpp_init_year  experiment_id   grid_label      institution_id  member_id       source_id       table_id        variable_id     version         zstore
julia> zopen(store.zstore[1])
ZarrGroup at GCStore("cmip6") and path CMIP6/ScenarioMIP/DKRZ/MPI-ESM1-2-HR/ssp585/r1i1p1f1/day/tas/gn/v20190710/
Variables: lat height time tas lat_bnds lon_bnds lon time_bnds 

julia> zopen(store.zstore[1])
ZarrGroup at GCStore("cmip6") and path CMIP6/ScenarioMIP/DKRZ/MPI-ESM1-2-HR/ssp585/r1i1p1f1/day/tas/gn/v20190710/
Variables: lat height time tas lat_bnds lon_bnds lon time_bnds 

julia> @time zopen(store.zstore[1])
  1.190920 seconds (6.51 k allocations: 692.555 KiB)
ZarrGroup at GCStore("cmip6") and path CMIP6/ScenarioMIP/DKRZ/MPI-ESM1-2-HR/ssp585/r1i1p1f1/day/tas/gn/v20190710/
Variables: lat height time tas lat_bnds lon_bnds lon time_bnds 

julia> c = Cube(store.zstore[1])
ERROR: ArgumentError: invalid index: nothing of type Nothing
Stacktrace:
  [1] to_index(i::Nothing)
    @ Base ./indices.jl:300
  [2] to_index(A::Vector{Any}, i::Nothing)
    @ Base ./indices.jl:277
  [3] to_indices
    @ ./indices.jl:333 [inlined]
  [4] to_indices
    @ ./indices.jl:325 [inlined]
  [5] getindex
    @ ./abstractarray.jl:1241 [inlined]
  [6] concatenatecubes(cl::Vector{Any}, cataxis::CategoricalAxis{String, :Variable, Vector{String}})
    @ YAXArrays.Cubes ~/.julia/packages/YAXArrays/Fe7F8/src/Cubes/TransformedCubes.jl:31
  [7] Cube(ds::Dataset; joinname::String)
    @ YAXArrays.Datasets ~/.julia/packages/YAXArrays/Fe7F8/src/DatasetAPI/Datasets.jl:324
  [8] Cube
    @ ~/.julia/packages/YAXArrays/Fe7F8/src/DatasetAPI/Datasets.jl:299 [inlined]
  [9] #Cube#122
    @ ~/.julia/packages/YAXArrays/Fe7F8/src/DatasetAPI/Datasets.jl:744 [inlined]
 [10] Cube(s::String)
    @ YAXArrays.Datasets ~/.julia/packages/YAXArrays/Fe7F8/src/DatasetAPI/Datasets.jl:744
 [11] top-level scope
    @ REPL[15]:1

Reproduce the failure with the CMIP6 dataset example, then inspect concatenatecubes in src/Cubes/TransformedCubes.jl and Cube in src/DatasetAPI/Datasets.jl, as shown in the stack trace. Determine how the zero-dimensional subcube reaches the invalid index operation. Done means the input produces a clear error explaining the unsupported zero-dimensional subcube instead of invalid index: nothing.