ARROW-8425: [Rust] [Parquet] Correct temporal IO

[nevi-me]

Contains the following:

• Fixes the date64 writer by writing the array as a timestamp milli
• Adds support for reading and writing interval types

Apache Spark doesn't read intervals (as they're fixed len binary)
It looks like the CPP impl at v2.0.0 just reads back the binary data.
The binary data read is however correct.

The result is that we can now correctly read and write the temporal types.
The Duration type remains unsupported.

[nevi-me]

CC @carols10cents @houqp

[codecov-io]

Codecov Report

Merging #8926 (c760e88) into master (091df20) will decrease coverage by 0.03%.
The diff coverage is 64.62%.

@@            Coverage Diff             @@
##           master    #8926      +/-   ##
==========================================
- Coverage   83.22%   83.18%   -0.04%     
==========================================
  Files         196      196              
  Lines       48232    48341     +109     
==========================================
+ Hits        40142    40214      +72     
- Misses       8090     8127      +37     

Impacted Files	Coverage Δ
rust/parquet/src/arrow/array_reader.rs	75.49% <48.43%> (-1.51%)	⬇️
rust/parquet/src/arrow/schema.rs	91.05% <56.52%> (-0.27%)	⬇️
rust/parquet/src/arrow/arrow_writer.rs	95.58% <78.57%> (-1.76%)	⬇️
rust/parquet/src/arrow/arrow_reader.rs	91.35% <100.00%> (+0.09%)	⬆️
rust/parquet/src/arrow/converter.rs	73.04% <100.00%> (+4.35%)	⬆️
rust/parquet/src/encodings/encoding.rs	95.24% <0.00%> (-0.20%)	⬇️
rust/arrow/src/compute/kernels/cast.rs	96.84% <0.00%> (+0.06%)	⬆️

---

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 091df20...c760e88. Read the comment docs.

[github-actions]

https://issues.apache.org/jira/browse/ARROW-8425

[jorgecarleitao]

I did not see any changes in the tests, which I would expect for a change in semantics. Aren't we not testing this yet, or how can I verify that it is now correct?

[nevi-me]

I did not see any changes in the tests, which I would expect for a change in semantics. Aren't we not testing this yet, or how can I verify that it is now correct?

There were existing tests, but they were ignored. So I removed the #[ignore] so they run.

I was writing date64 as timestampmilli, but this wasn't per the spec, so I fixed that now. I also added an interval test to increase test coverage.

[jorgecarleitao]

LGTM.