Data type validation at DataFrame Models initialization seems to be broken in v0.21.0 compared to v0.20.4

[hnomichith]

Describe the bug

After upgrading from Pandera 0.20.4 to 0.21.0, the data type validation at DataFrame Models initialization (cf this doc) seems to be broken: validating a wrong dataframe does not raise an error.

• 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.

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

    import pandas as pd
    from pandera import DataFrameModel
    from pandera.typing import DataFrame

    class ExampleModel(DataFrameModel):
        field1: str
        field2: float

    # This is supposed to crash, the schema is not respected.
    # This is raising validation error (as expected) in 0.20.4, but not in 0.21.0
    dataframe = DataFrame[ExampleModel](
        pd.DataFrame.from_dict(
            {
                "field3": [None, None, 1, None, 2],
            }
        )
    )

    # Only reached in 0.21.0
    print(dataframe)

Expected behavior

The data type validation should raise an error, and the line print(dataframe) shouldnot be reached.

Desktop (please complete the following information):

• OS: Linux Mint
• Version: 21.2

[hasanasim]

Data type validation is also broken on v0.21.0 when using the pa.check decorator and works fine on v0.20.4 e.g.

class MyModel(pa.DataFrameModel):
    my_type: object
    id: int

    @pa.check("my_type")
    def is_serialised_type(cls, t: object) -> bool:
        return isinstance(t, SerialisedType)