Support copying from glob patterns

[aykut-bozkurt]

• Adds udf parquet.list(<uri_with_pattern>) where uri_with_pattern might contain * for words of arbitrary length or ** for arbitrarily nested directories.
• Support COPY FROM <uri_with_pattern>.

COPY test1 TO '/tmp/parent/child/test1.parquet';
COPY test2 TO '/tmp/parent/child/test2.parquet';

COPY test3 FROM '/tmp/parent/**/*.parquet' WITH (format 'parquet');

Warning: list operation is not supported for http(s) object stores. (available for all other stores)

Closes #112.

[codecov]

Codecov Report

❌ Patch coverage is 98.31461% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.30%. Comparing base (7060d99) to head (87f3d04).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/arrow_parquet/parquet_reader.rs	96.47%	5 Missing ⚠️
src/arrow_parquet/uri_utils.rs	89.47%	2 Missing ⚠️
src/parquet_udfs/list.rs	96.49%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #120      +/-   ##
==========================================
+ Coverage   91.03%   91.30%   +0.27%     
==========================================
  Files          95       97       +2     
  Lines       10759    11204     +445     
==========================================
+ Hits         9794    10230     +436     
- Misses        965      974       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[pgguru]

So what happens if the files have different structures and they're non-castable/other columns?

I see the check for the schema matching the tuple type, so I assume it'd presumably start loading things until it hit a file that couldn't be coerced and abort at that point. Is it worth adding a test of different structures to validate behavior here?

[pgguru]

Pedantically, anything that contains ** also contains *. Is it possible to have escaped * chars that aren't pattern chars, or other glob syntaxes (like % or ? or something)?

[aykut-bozkurt]

added one test with s3://.../** in addition to s3://.../**/*.parquet. I did not want to add support for % or ? initially. (a bit complicated with rust pattern library)

[aykut-bozkurt]

hmm what happens if * is allowed as file or uri key. let me check.

[aykut-bozkurt]

flow is like below:

1. try reading file from uri
2. if the uri does not exist
   2.1 if uri contains pattern (only '*' for now), read after listing the uri
   2.2 otherwise panic with original error

[aykut-bozkurt]

added tests with schema mismatch (more columns, or column type mismatch) Those are checked before reading any file.

[pgguru]

Looks good with changes, thanks!