feat(python): Raise DuplicateError if given a pyarrow Table object with duplicate column names

[alexander-beedie]

Closes #20616.

We already catch this case in read_database (see: #18548), but read_database_uri has exposed a more generic gremlin in frame construction from pyarrow.Table objects (eg: connectorx loads query results to pyarrow.Table, which gets passed to pl.from_arrow).

We weren't explicitly checking the incoming pyarrow.Table for duplicate columns, which leads to a somewhat non-obvious ShapeError; with this PR we now raise a more helpful DuplicateError that explicitly refers to the column name that appears more than once.

Example

import pyarrow as pa
import polars as pl

tbl = pa.Table.from_arrays(
    arrays=[
        pa.array([1, 2, 3], type=pa.int32()),
        pa.array([4, 5, 6], type=pa.int32()),
    ],
    schema=pa.schema([
        ("dupe_col", pa.int32()), 
        ("dupe_col", pa.int32()),
    ]),
)

🔴 Before

pl.from_arrow(tbl)
# ShapeError: 2 column names provided for a DataFrame of width 1

🟢 After

pl.from_arrow(tbl)
# DuplicateError: column 'dupe_col' appears 2 times; names must be unique

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.04%. Comparing base (92fd75d) to head (9d1cba4).
Report is 15 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #20624      +/-   ##
==========================================
+ Coverage   79.00%   79.04%   +0.04%     
==========================================
  Files        1566     1557       -9     
  Lines      221034   220543     -491     
  Branches     2510     2514       +4     
==========================================
- Hits       174618   174325     -293     
+ Misses      45842    45644     -198     
  Partials      574      574              

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