[Containers] convert axes to AbstractVector in DenseAxisArray

[odow]

This PR fixes some confusing behavior with iterative non AbstractVector axes in DenseAxisArray.

Closes #2424

Note that now the axis is a vector of pairs, not a dictionary (which guarantees replicable iteration order), and you can access via tuples or pairs.

Before

julia> d = Dict("a" => 1, "b" => 2)
Dict{String,Int64} with 2 entries:
  "b" => 2
  "a" => 1

julia> model = Model();

julia> @variable(model, x[(k, v) in d])
1-dimensional DenseAxisArray{VariableRef,1,...} with index sets:
    Dimension 1, Dict("b" => 2,"a" => 1)
And data, a 2-element Array{VariableRef,1}:
 x[("b", 2)]
 x[("a", 1)]

julia> x[("a", 1)]
ERROR: KeyError: key ("a", 1) not found
Stacktrace:
 [1] getindex at ./dict.jl:467 [inlined]
 [2] lookup_index at /Users/oscar/.julia/packages/JuMP/qhoVb/src/Containers/DenseAxisArray.jl:140 [inlined]
 [3] _to_index_tuple at /Users/oscar/.julia/packages/JuMP/qhoVb/src/Containers/DenseAxisArray.jl:149 [inlined]
 [4] to_index at /Users/oscar/.julia/packages/JuMP/qhoVb/src/Containers/DenseAxisArray.jl:167 [inlined]
 [5] getindex(::JuMP.Containers.DenseAxisArray{VariableRef,1,Tuple{Dict{String,Int64}},Tuple{Dict{Pair{String,Int64},Int64}}}, ::Tuple{String,Int64}) at /Users/oscar/.julia/packages/JuMP/qhoVb/src/Containers/DenseAxisArray.jl:182
 [6] top-level scope at REPL[68]:1

julia> x["a" => 1]
x[("a", 1)]

Now

julia> using JuMP

julia> d = Dict("a" => 1, "b" => 2)
Dict{String, Int64} with 2 entries:
  "b" => 2
  "a" => 1

julia> model = Model();

julia> @variable(model, x[(k, v) in d])
1-dimensional DenseAxisArray{VariableRef,1,...} with index sets:
    Dimension 1, ["b" => 2, "a" => 1]
And data, a 2-element Vector{VariableRef}:
 x[("b", 2)]
 x[("a", 1)]

julia> x["a" => 1]
x[("a", 1)]

julia> x[("a", 1)]
x[("a", 1)]

[codecov]

Codecov Report

Merging #2713 (f90a575) into master (4f0ecb4) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #2713   +/-   ##
=======================================
  Coverage   93.62%   93.62%           
=======================================
  Files          44       44           
  Lines        5455     5460    +5     
=======================================
+ Hits         5107     5112    +5     
  Misses        348      348           

Impacted Files	Coverage Δ
src/Containers/DenseAxisArray.jl	88.77% <100.00%> (+0.30%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4f0ecb4...f90a575. Read the comment docs.