Improved inference and deserialization of CSV

[jorgecarleitao]

This PR:

• improves performance of inference by removing the &[u8] -> utf8 for types that do not need such conversion (integer, boolean, float)
• Added support to infer and deserialize timestamps with timezone
• Added support to deserialize all timestamp units
• Removed two dependencies to reading csv: lazy_static and regex, so that the inference is 100% consistent with deserialization

The deserialization and inference of timestamps with timezone assumes RFC3339. Users may extend the inferer and deserializer to handle other formats.

This also fixes a bug on which datetimes were being parsed as Date64, when date64 is expected to be a multiple of 86400000 and not an arbitrary number. They are now inferer and parsed as a Timestamp(Millisecond, None)

[codecov]

Codecov Report

Merging #483 (8a90ece) into main (c098494) will decrease coverage by 0.02%.
The diff coverage is 78.78%.

@@            Coverage Diff             @@
##             main     #483      +/-   ##
==========================================
- Coverage   80.28%   80.25%   -0.03%     
==========================================
  Files         374      374              
  Lines       22949    23753     +804     
==========================================
+ Hits        18424    19063     +639     
- Misses       4525     4690     +165     

Impacted Files	Coverage Δ
src/io/csv/read/reader.rs	47.36% <ø> (-15.79%)	⬇️
src/io/csv/read/deserialize.rs	63.86% <59.25%> (+6.43%)	⬆️
src/io/csv/read/infer_schema.rs	85.71% <81.48%> (+0.42%)	⬆️
tests/it/io/csv/read.rs	100.00% <100.00%> (ø)
src/compute/cast/binary_to.rs	71.73% <0.00%> (-4.13%)	⬇️
src/compute/take/mod.rs	74.57% <0.00%> (-1.19%)	⬇️
src/io/avro/read/schema.rs	41.37% <0.00%> (-1.15%)	⬇️
src/temporal_conversions.rs	74.54% <0.00%> (-0.81%)	⬇️
src/compute/comparison/simd/mod.rs	95.12% <0.00%> (-0.72%)	⬇️
src/types/simd/native.rs	90.32% <0.00%> (-0.59%)	⬇️
... and 22 more

---

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 c098494...8a90ece. Read the comment docs.