Support Vector{Dict{String,Any}} as a Table format

[robsmith11]

I see that support was added from #132 but it looks like it's only working for the case where keys are Symbols, not Strings, which for example is what JSON.Parse.parse() outputs.

Is there any reason to not support both cases?

julia> DataFrame([Dict(:x => rand(), :y => rand(Int)) for _ in 1:5])
5×2 DataFrame
 Row │ y                     x
     │ Int64                 Float64
─────┼─────────────────────────────────
   1 │ -6168923119116954553  0.733732
   2 │ -5238086163398295723  0.684467
   3 │  7996475788962267833  0.88606
   4 │ -1641761153302011089  0.0740284
   5 │  8099478073091371152  0.980377

julia> DataFrame([Dict("x" => rand(), "y" => rand(Int)) for _ in 1:5])
5×8 DataFrame
 Row │ slots                              keys                               vals                               ndel   count  age     idxfloor  maxprobe
     │ Array…                             Array…                             Array…                             Int64  Int64  UInt64  Int64     Int64
─────┼───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
   1 │ UInt8[0x00, 0x00, 0x00, 0x00, 0x…  [#undef, #undef, #undef, #undef,…  Real[#undef, #undef, #undef, #un…      0      2       2         9         0
   2 │ UInt8[0x00, 0x00, 0x00, 0x00, 0x…  [#undef, #undef, #undef, #undef,…  Real[#undef, #undef, #undef, #un…      0      2       2         9         0
   3 │ UInt8[0x00, 0x00, 0x00, 0x00, 0x…  [#undef, #undef, #undef, #undef,…  Real[#undef, #undef, #undef, #un…      0      2       2         9         0
   4 │ UInt8[0x00, 0x00, 0x00, 0x00, 0x…  [#undef, #undef, #undef, #undef,…  Real[#undef, #undef, #undef, #un…      0      2       2         9         0
   5 │ UInt8[0x00, 0x00, 0x00, 0x00, 0x…  [#undef, #undef, #undef, #undef,…  Real[#undef, #undef, #undef, #un…      0      2       2         9         0

[bkamins]

AFAICT the thing is that normally assumes that column names are Symbol, but my opinion is that this case is so common that it could get a special handling. If @quinnj rejects this in Tables.jl we can discuss of adding a special constructor for this case in DataFrames.jl like now we have for AbstractDict{String} as this is the same case:

julia> DataFrame(Dict("A" => 1:2, "B" => 2:3))
2×2 DataFrame
 Row │ A      B
     │ Int64  Int64
─────┼──────────────
   1 │     1      2
   2 │     2      3

julia> Tables.columntable(Dict("A" => 1, "B" => 2:3))
ERROR: to treat Dict{String, Any} as a table, it must have a key type of `Symbol`, and a value type `<: AbstractVector`

[quinnj]

Implemented in #296

[bkamins]

@robsmith11 - OK to close it?

[robsmith11]

Yep, just tested master branch and works great now. Thanks!