Move @df macro to Plots

[daschw]

fix #3336

[codecov]

Codecov Report

Merging #3351 (ebf9116) into master (060162c) will decrease coverage by 0.99%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #3351      +/-   ##
==========================================
- Coverage   64.81%   63.81%   -1.00%     
==========================================
  Files          26       27       +1     
  Lines        6539     6639     +100     
==========================================
- Hits         4238     4237       -1     
- Misses       2301     2402     +101     

Impacted Files	Coverage Δ
src/Plots.jl	34.28% <ø> (ø)
src/df.jl	0.00% <0.00%> (ø)
src/examples.jl	61.19% <ø> (-0.58%)	⬇️

---

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 060162c...2d6f43f. Read the comment docs.

[isentropic]

Perhaps this code should instead go to recipes.jl?

[mkborregaard]

Nice 👍

[getzze]

Is there anything in the code that is specific to plotting? Otherwise I would say this should be defined in Tables.jl, no?
I always have a hard time remembering which package imports @df, as there is nothing in the name or the functionality that makes me think of Plots (or StatsPlots).

[Ickaser]

It's been a while since any discussion happened on this; to summarize the discussion at JuliaData/Tables.jl#287, it looks like they already have a macro @with that is "better" but relies on some DataFrames functionality so that it is not obvious how to implement for general tables, as works for the @df macro. That being the case, I as a user would be in favor of moving @df to Plots, since that is the only thing that I use from StatsPlots. I have been using it for TypedTables, rather than DataFrames, so if it were to be renamed to @table (or something similar) I would be on board.