Scalar return for check in polars-backed model fails on validation with lazy=True

[SandroCasagrande]

Describe the bug
A clear and concise description of what the bug is.

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of pandera.
• (optional) I have confirmed this bug exists on the main branch of pandera.

Code Sample, a copy-pastable example

I am using scalar returns for checks on a polars-backed DataFrameModel

import pandera.polars as pa
import polars as pl

class MyModel(pa.DataFrameModel):
    a: pl.Int64

    @pa.check("a")
    def failing_check(self, _) -> bool:
        return False

MyModel(pl.DataFrame({"a": [1]}), lazy=True)

which results in the following traceback

Traceback (most recent call last):
  File "/Users/sandro/code/pandera/scalar_returns_polars.py", line 12, in <module>
    MyModel(pl.DataFrame({"a": [1]}), lazy=True)
  File "/Users/sandro/code/pandera/pandera/api/dataframe/model.py", line 138, in __new__
    DataFrameBase[TDataFrameModel], cls.validate(*args, **kwargs)
                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/api/dataframe/model.py", line 289, in validate
    cls.to_schema().validate(
  File "/Users/sandro/code/pandera/pandera/api/polars/container.py", line 58, in validate
    output = self.get_backend(check_obj).validate(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/backends/polars/container.py", line 89, in validate
    results = check(*args)  # type: ignore[operator]
              ^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/backends/polars/container.py", line 179, in run_schema_component_checks
    result = schema_component.validate(check_obj, lazy=lazy)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/api/polars/components.py", line 146, in validate
    output = self.get_backend(check_obj).validate(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/backends/polars/components.py", line 80, in validate
    error_handler = self.run_checks_and_handle_errors(
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sandro/code/pandera/pandera/backends/polars/components.py", line 146, in run_checks_and_handle_errors
    error_handler.collect_error(
  File "/Users/sandro/code/pandera/pandera/api/base/error_handler.py", line 69, in collect_error
    else len(schema_error.failure_cases)
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: object of type 'bool' has no len()

Expected behavior

A specific SchemaErrors should be raise, like it happens with an analogous pandas-backed model

import pandas as pd
import pandera as pa

class MyModel(pa.DataFrameModel):
    a: pd.Int64Dtype

    @pa.check("a")
    def failing_check(self, _) -> bool:
        return False

MyModel(pd.DataFrame({"a": [1]}), lazy=True)

which results in

Traceback (most recent call last):
...
pandera.errors.SchemaErrors: {
    "DATA": {
        "DATAFRAME_CHECK": [
            {
                "schema": "MyModel",
                "column": "a",
                "check": "failing_check",
                "error": "Column 'a' failed series or dataframe validator 0: <Check failing_check>"
            }
        ]
    }
}

Desktop (please complete the following information):

• pandera version: 0.19.3 and c24dda9
• polars version: 0.20.30

Additional context

There is an obvious difference between the handling of scalar False values in the polars backend vs the pandas backend. In the former failure_case remains scalar, which leads to the error when counting failure cases (using lazy=True). Applying an analogous wrapping like scalar_failure_case for pandas seems to work in the very simple case above, but I'm not sure if this approach holds for anything beyond my example.