feat: Add SQL support for GROUP BY ALL syntax and fix several issues with aliased group keys

[alexander-beedie]

Closes #15585 (ref: #14704 (comment))

Feature

GROUP BY ALL is a useful SQL construct (as implemented by BigQuery, DuckDB, Snowflake, etc) that allows for automatic grouping by column expressions in the SELECT clause that are not (and do not contain) aggregate/window functions.

from datetime import date
import polars as pl

dt1 = date(1999, 12, 31)
dt2 = date(2028, 7, 5)

df = pl.DataFrame({
    "key": ["xx", "yy", "xx", "yy", "xx", "xx"],
    "dt": [dt1, dt1, dt1, dt2, dt2, dt2],
    "value": [10.5, -5.5, 20.5, 8.0, -3.0, 5.0],
})

df.sql(
    """
    SELECT dt, key, SUM(value) AS total,
    FROM self
    GROUP BY ALL
    ORDER BY dt, key
    """
)
# shape: (4, 3)
# ┌────────────┬─────┬───────┐
# │ dt         ┆ key ┆ total │
# │ ---        ┆ --- ┆ ---   │
# │ date       ┆ str ┆ f64   │
# ╞════════════╪═════╪═══════╡
# │ 1999-12-31 ┆ xx  ┆ 31.0  │
# │ 1999-12-31 ┆ yy  ┆ -5.5  │
# │ 2028-07-05 ┆ xx  ┆ 2.0   │
# │ 2028-07-05 ┆ yy  ┆ 8.0   │
# └────────────┴─────┴───────┘

Fixes

• While implementing this I came across several errors with group-key aliasing/tracking, resolved them, and added test coverage. I'm aware of one more similar gremlin, but it's very much an edge-case and I'll address it in a subsequent PR.
• Use of ARRAY_AGG in conjunction with GROUP BY could result in a double-implode, due to our implicit array aggregation on bare agg columns in the regular API.
• Attempting to group by ordinal values should have raised on negative integers - and now it does.

[codecov]

Codecov Report

Attention: Patch coverage is 95.50562% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 80.99%. Comparing base (3892c75) to head (b3231ea).
Report is 3 commits behind head on main.

Files	Patch %	Lines
crates/polars-sql/src/context.rs	94.66%	4 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #16179   +/-   ##
=======================================
  Coverage   80.99%   80.99%           
=======================================
  Files        1387     1392    +5     
  Lines      178832   178920   +88     
  Branches     2877     2892   +15     
=======================================
+ Hits       144844   144923   +79     
- Misses      33496    33500    +4     
- Partials      492      497    +5     

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