Support by/by_left/by_right arguments in join_where

[mcrumiller]

Description

For join_asof, we can currently supply columns to do a "regular" join on, in addition to the as-of columns to join. In a join_where, we must explicitly supply the equality conditions on all of the extra fields, which I assume (maybe incorrectly) takes a less efficient path than a regular join would on those columns:

import polars as pl
from polars import col

df1 = pl.DataFrame({
    "a": [1, 1, 2, 2],
    "b": [1, 2, 3, 4],
})
df2 = pl.DataFrame({
    "a": [1, 2, 2, 3],
    "c": [1, 2, 5, 6],
})

# Current syntax: explicitly provide equality columns
df1.join_where(
    df2,
    col("b") < col("c"),
    col("a") == col("a_right"),
)
# shape: (2, 3)
# ┌─────┬─────┬─────┐
# │ a   ┆ b   ┆ c   │
# │ --- ┆ --- ┆ --- │
# │ i64 ┆ i64 ┆ i64 │
# ╞═════╪═════╪═════╡
# │ 2   ┆ 3   ┆ 5   │
# │ 2   ┆ 4   ┆ 5   │
# └─────┴─────┴─────┘

# Proposed syntax: use by
df1.join_where(
    df2,
    col("b") < col("c"),
    by="a",  # alternative (and can provide list)
)