fix(python): Address incorrect align_frames result when the alignment column contains NULL values

[alexander-beedie]

Internally align_frames uses joins, but the default behaviour for joins is for NULL values not to match, which can result in extraneous/incorrect rows being generated during frame alignment if the alignment column(s) contain such values.

This PR fixes that oversight 👌

Example

import polars as pl

df1 = pl.DataFrame({
  "key": ["x", "y", None],
  "value": [1, 2, 0],
})
df2 = pl.DataFrame({
  "key": ["x", None, "z", "y"],
  "value": [4, 3, 6, 5],
})

# ┌──────┬───────┐   ┌──────┬───────┐   
# │ key  ┆ value │   │ key  ┆ value │   
# │ ---  ┆ ---   │   │ ---  ┆ ---   │   
# │ str  ┆ i64   │   │ str  ┆ i64   │   
# ╞══════╪═══════╡   ╞══════╪═══════╡   
# │ x    ┆ 1     │   │ x    ┆ 4     │   
# │ y    ┆ 2     │   │ null ┆ 3     │   
# │ null ┆ 0     │   │ z    ┆ 6     │   
# └──────┴───────┘   │ y    ┆ 5     │   
#                    └──────┴───────┘   

Before

Additional row gets (incorrectly) introduced:

pl.align_frames(df1, df2, on="key")
# ┌──────┬───────┐   ┌──────┬───────┐
# │ key  ┆ value │   │ key  ┆ value │
# │ ---  ┆ ---   │   │ ---  ┆ ---   │
# │ str  ┆ i64   │   │ str  ┆ i64   │
# ╞══════╪═══════╡   ╞══════╪═══════╡
# │ null ┆ null  │   │ null ┆ 3     │
# │ null ┆ 0     │   │ null ┆ null  │
# │ x    ┆ 1     │   │ x    ┆ 4     │
# │ y    ┆ 2     │   │ y    ┆ 5     │
# │ z    ┆ null  │   │ z    ┆ 6     │
# └──────┴───────┘   └──────┴───────┘

After

Correct number of appropriately aligned rows:

pl.align_frames(df1, df2, on="key")
# ┌──────┬───────┐   ┌──────┬───────┐
# │ key  ┆ value │   │ key  ┆ value │
# │ ---  ┆ ---   │   │ ---  ┆ ---   │
# │ str  ┆ i64   │   │ str  ┆ i64   │
# ╞══════╪═══════╡   ╞══════╪═══════╡
# │ null ┆ 0     │   │ null ┆ 3     │
# │ x    ┆ 1     │   │ x    ┆ 4     │
# │ y    ┆ 2     │   │ y    ┆ 5     │
# │ z    ┆ null  │   │ z    ┆ 6     │
# └──────┴───────┘   └──────┴───────┘

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.86%. Comparing base (ab25b3e) to head (e69118e).
Report is 6 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #18521   +/-   ##
=======================================
  Coverage   79.85%   79.86%           
=======================================
  Files        1501     1501           
  Lines      201829   201829           
  Branches     2868     2868           
=======================================
+ Hits       161174   161187   +13     
+ Misses      40109    40096   -13     
  Partials      546      546           

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