Improved performance and stability of writing to CSV

[ritchie46]

This addresses #865. It is a breaking change, but it gives us more control and very likely improves performance because we only use csv-core for utf8 fields.

Because we have a lot less variants than the original csv crate has to deal with. E.g.

• we know the schema,
• we know that we will have an equal number of rows and fields

[codecov]

Codecov Report

Merging #866 (bbd900d) into main (10e6cd5) will decrease coverage by 0.03%.
The diff coverage is 60.78%.

@@            Coverage Diff             @@
##             main     #866      +/-   ##
==========================================
- Coverage   71.56%   71.52%   -0.04%     
==========================================
  Files         335      335              
  Lines       17952    17989      +37     
==========================================
+ Hits        12847    12867      +20     
- Misses       5105     5122      +17     

Impacted Files	Coverage Δ
src/io/csv/write/mod.rs	57.77% <50.00%> (-7.61%)	⬇️
src/io/csv/write/serialize.rs	60.16% <76.19%> (-0.61%)	⬇️
src/compute/arithmetics/time.rs	25.68% <0.00%> (-0.92%)	⬇️
src/io/json/write/serialize.rs	86.11% <0.00%> (-0.22%)	⬇️
src/io/json/write/mod.rs	100.00% <0.00%> (ø)

---

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 10e6cd5...bbd900d. Read the comment docs.

[ritchie46]

Still have to do the parallel example.

[jorgecarleitao]

Great PR!

Indeed this will be more performant. I left one comment.

[ritchie46]

It should be good to go.

[jorgecarleitao]

Indeed. Thanks a lot, @ritchie46 !