add shorter show method for AbstractColumns

[ericphanson]

For example,

julia> big_tbl = (; a = [rand(100) for _ = 1:200], b = [rand(100) for _ = 1:200], c = [rand(100) for _ = 1:200]);

julia> tbl = Arrow.Table(Arrow.tobuffer(big_tbl))
Arrow.Table with 200 rows, 3 columns, and schema:
 :a  Vector{Float64} (alias for Array{Float64, 1})
 :b  Vector{Float64} (alias for Array{Float64, 1})
 :c  Vector{Float64} (alias for Array{Float64, 1})

This table used to print as 1.2M characters! (The “alias” printing is removed with JuliaLang/julia#39256 I believe).

I tried to be pretty safe here since AbstractColumns are very general, and I'm not sure if we can expect columns to have a length for example.

I decided to print the schema instead of a manual column printing as in apache/arrow-julia#168, since we don't have the eltype info in general (columntypes just asks the schema, so the fallback method only works when we have a schema), and I think maybe if Tables doesn't have a schema that can be useful info to pass along to the user (perhaps they can add one and improve performance in some cases).

[codecov]

Codecov Report

Merging #236 (212777c) into main (8ed01b2) will increase coverage by 0.07%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #236      +/-   ##
==========================================
+ Coverage   94.51%   94.59%   +0.07%     
==========================================
  Files           7        7              
  Lines         602      610       +8     
==========================================
+ Hits          569      577       +8     
  Misses         33       33              

Impacted Files	Coverage Δ
src/Tables.jl	87.50% <100.00%> (+0.78%)	⬆️

---

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 8ed01b2...212777c. Read the comment docs.

[quinnj]

Thanks @ericphanson !