fix: PandasPreprocessor._spark_apply_steps breaks due to bad kwarg

[Lodewic]

Fixes #46

The PandasPreprocessor._spark_apply_steps is called when:

1. Using PandasEntityMatching
2. passing spark_session to the model
3. processing 200k+ records

But, it breaks because a kwarg is passed with the wrong name.

Reproduce the problem

In this PR, this gets fixed.

import pandas as pd  # pandas 2.3.3
from pyspark.sql import SparkSession  # pyspark 3.5.2

from emm import PandasEntityMatching  # emm 2.1.10

spark = SparkSession.builder.getOrCreate()

# If spark_session is passed, and at least 200k records processed, 
# then Spark will be used for transformations. Otherwise, pandas will be used.
model = PandasEntityMatching(parameters={"spark_session": spark})

# Build 200k+ records
names = [
    ('John Smith LLC'),
    ('ING LLC'),
    ('John Doe LLC'),
    ('Zhe Sun G.M.B.H'),
    ('Random GMBH'),
] * 100_000
df = pd.DataFrame(names, columns=['name']).assign(id=range(len(names)))

# This will raise an error.
model.fit(df) # >>> TypeError: PandasPreprocessor._spark_apply_steps.<locals>.calc() got an unexpected keyword argument 'functions'

The fix

Fix the named keyword argument in:

def calc(chunk, funcs):
    for func in funcs:
        chunk = func(chunk)
     return chunk.index.values, chunk.values

to

def calc(chunk, functions):
    for func in functions:
        chunk = func(chunk)
     return chunk.index.values, chunk.values

this matches how the function is actually called.

Tests

I suppose there are no tests for this already, so I also have not updated or added any.

[mbaak]

Thanks!