fix: Remove assert that panics on group_by followed by head(n), where n is larger then the frame height

[alexander-beedie]

Closes #20818.

Following the code further down, I think the panicking assert is invalid, and can be removed... @ritchie46, could you double-check, in case I missed some other paths that do rely on this?🤞😅 It's not an error to ask for a slice larger than the number of available rows - in this case it should just get clipped to how many rows there actually are.

Example

The 1.20.0 release panics on the following:

import polars as pl

lf = pl.LazyFrame({
    "key": ["xx", "yy", "zz", "xx", "zz", "zz"],
    "val": [15, 25, 10, 20, 20, 20],
})
grp = lf.group_by("key").agg(pl.col("val").sum())
grp.head(5).collect()  # aggregate result is 3 rows; head(5) panics

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.69%. Comparing base (099ee3c) to head (2378854).
Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #20819      +/-   ##
==========================================
+ Coverage   79.63%   79.69%   +0.06%     
==========================================
  Files        1568     1568              
  Lines      222970   222976       +6     
  Branches     2544     2544              
==========================================
+ Hits       177555   177699     +144     
+ Misses      44831    44693     -138     
  Partials      584      584              

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

[ritchie46]

The length should then be truncated to the len of the groups. It is in invariant of the GroupsPositions that the len is in bounds.

[alexander-beedie]

The length should then be truncated to the len of the groups. It is in invariant of the GroupsPositions that the len is in bounds.

Yup; seemed to be happening implicitly somewhere - have made the truncation explicit, and added a small note about the invariant 👍