Support Expr.assert_all_true expression, which preserves current values but raises error at runtime if any value is False.

[cosmicBboy]

Description

Hi,

I'm the maintainer of pandera, and we recently released support for validating polars LazyFrames and DataFrames: https://pandera.readthedocs.io/en/latest/polars.html

One caveat is that there's different behavior between LazyFrame and DataFrame validation: with the former, it only validates schema-level properties (columns and datatypes) because we don't want to materialize any data on a schema.validate call (e.g. here).

@register_builtin_check(
    aliases=["eq"],
    error="equal_to({value})",
)
def equal_to(data: PolarsData, value: Any) -> pl.LazyFrame:
    """Ensure all elements of a data container equal a certain value.

    :param data: NamedTuple PolarsData contains the dataframe and column name for the check. The keys
                to access the dataframe is "dataframe" and column name using "key".
    :param value: values in this polars data structure must be
        equal to this value.
    """
    return data.lazyframe.select(pl.col(data.key).eq(value))

We designed this to be forward-looking, and the expected signature of a data-level check should return LazyFrames and leverage the lazy API as much as possible. I think a great next iteration for this integration would be great to be able to do data-level validations that preserves the values of the expression but raises an error at runtime if any value is False:

def equal_to(data: PolarsData, value: Any) -> pl.LazyFrame:
    return data.lazyframe.select(pl.col(data.key).eq(value).assert_all_true())

Context: this was suggested by @MarcoGorelli here: https://discord.com/channels/897120336003334214/1160076741394563102/1237738610837557289