[FEAT] UTF8 to binary coercion flag

[raunakab]

Overview

• add string-to-binary coercion flag
• cleaned up some logic and instances of non-idiomatic Rust code

[codspeed-hq]

CodSpeed Performance Report

Merging #2893 will not alter performance

_{Comparing feature/utf8-to-binary-coercion (58c0730) with main (b519944)}

Summary

✅ 17 untouched benchmarks

[codecov]

Codecov Report

Attention: Patch coverage is 88.14815% with 16 lines in your changes missing coverage. Please review.

Project coverage is 66.16%. Comparing base (78a92a2) to head (58c0730).
Report is 23 commits behind head on main.

Files with missing lines	Patch %	Lines
src/daft-parquet/src/read.rs	87.80%	10 Missing ⚠️
src/arrow2/src/io/parquet/read/schema/mod.rs	53.84%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2893      +/-   ##
==========================================
+ Coverage   65.93%   66.16%   +0.23%     
==========================================
  Files        1010     1002       -8     
  Lines      113440   114108     +668     
==========================================
+ Hits        74794    75497     +703     
+ Misses      38646    38611      -35     

Flag	Coverage Δ
	66.16% <88.14%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
daft/table/table.py	61.01% <100.00%> (ø)
src/arrow2/src/io/parquet/read/schema/convert.rs	95.88% <100.00%> (+0.04%)	⬆️
src/daft-micropartition/src/micropartition.rs	90.75% <100.00%> (-2.07%)	⬇️
src/daft-parquet/src/file.rs	74.20% <100.00%> (+0.03%)	⬆️
src/daft-parquet/src/lib.rs	50.00% <ø> (ø)
src/daft-parquet/src/python.rs	58.91% <100.00%> (+0.15%)	⬆️
src/daft-parquet/src/stream_reader.rs	90.56% <100.00%> (+0.20%)	⬆️
src/daft-scan/src/glob.rs	87.64% <100.00%> (+0.04%)	⬆️
src/arrow2/src/io/parquet/read/schema/mod.rs	76.47% <53.84%> (-10.49%)	⬇️
src/daft-parquet/src/read.rs	74.66% <87.80%> (+1.19%)	⬆️

... and 175 files with indirect coverage changes

[jaychia]

Needs tests!

[jaychia]

Why do we need a builder? Could we not just implement a ParquetSchemaInferenceOptions::new()?

[raunakab]

We could, but the new function will seem to get busy in the future... I don't mind adding another param to ParquetSchemaInferenceOptions::new, but as we continue to add more features and flags to those options, seems like we'll end up deferring back to a builder anyways.

[jaychia]

Stamped with remaining comments