Constructor interpartition_merges & fixes/small improvements

[krynju]

WIP: trying out some things on constructors to make them faster/more reliable

Findings/features

1. materialization before ingesting the input avoids the huge allocation numbers that scale with input length

# pre change
julia> @time DTable(CSV.File("test.csv"), 12000)

  0.141167 seconds (4.92 M allocations: 123.415 MiB)
DTable with 84 partitions
Tabletype: NamedTuple

# post change

julia> @time DTable(CSV.File("test.csv"), 12000)

  0.021110 seconds (8.09 k allocations: 48.421 MiB)
DTable with 84 partitions
Tabletype: NamedTuple

2. interpartition merges

julia> DTable(CSV.Chunks("test.csv"), 70_000) |> DTables.chunk_lengths
15-element Vector{Int64}:
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 70000
 20000

julia> DTable(CSV.Chunks("test.csv"), 70_000, interpartition_merges=false) |> DTables.chunk_lengths
16-element Vector{Int64}:
 62495
 62502
 62505
 62493
 62505
 62501
 62492
 62507
 62495
 62502
 62505
 62493
 62505
 62501
 62492
 62507

julia> DTable(CSV.Chunks("test.csv"), 50_000) |> DTables.chunk_lengths
20-element Vector{Int64}:
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000
 50000

julia> DTable(CSV.Chunks("test.csv"), 50_000, interpartition_merges=false) |> DTables.chunk_lengths
32-element Vector{Int64}:
 50000
 12495
 50000
 12502
 50000
 12505
 50000
 12493
 50000
 12505
 50000
 12501
     ⋮
 12505
 50000
 12493
 50000
 12505
 50000
 12501
 50000
 12492
 50000
 12507

3. getting columns throgh properties works now and returns a lazy DTableColumn iterator
   it can be used for efficiently iterating through the column and it has a specialized collect function defined that uses the Tables.getcolumn implementation that efficiently links all column pieces into a SentinelArray

julia> d.a |> collect
100000-element SentinelArrays.ChainedVector{Int64, Vector{Int64}}:
 1
 2
 0
 1
 2
 0
 1
 ⋮
 1
 2
 0
 1
 2
 0
 1

julia> d.a |> sum
100000

3. docs updates

[codecov-commenter]

Codecov Report

Base: 90.25% // Head: 92.49% // Increases project coverage by +2.23% 🎉

Coverage data is based on head (9472c23) compared to base (dea0da6).
Patch coverage: 96.92% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #13      +/-   ##
==========================================
+ Coverage   90.25%   92.49%   +2.23%     
==========================================
  Files          10       10              
  Lines         852      879      +27     
==========================================
+ Hits          769      813      +44     
+ Misses         83       66      -17     

Impacted Files	Coverage Δ
src/table/dtable_column.jl	97.50% <66.66%> (-2.50%)	⬇️
src/table/dtable.jl	93.27% <98.07%> (+9.23%)	⬆️
src/table/dataframes_interface_utils.jl	92.72% <100.00%> (ø)
src/table/gdtable.jl	76.27% <100.00%> (ø)
src/table/operations.jl	98.48% <100.00%> (ø)
src/table/tables.jl	100.00% <100.00%> (+11.95%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.