Nullable pa.Date field with add_missing_columns=True fails validation when column is auto-added with pd.NaT default

[bBekan-Maistra]

Bug Description

When using add_missing_columns = True with a pa.Date field that has nullable=True and a default value of pd.NaT, Pandera raises a SchemaError claiming the series doesn't have type date, even though the field is properly configured to accept null values.

• 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:

import pandas as pd
import pandera as pa

class BaseValidationSchema(pa.DataFrameModel):
    class Config:
        add_missing_columns = True
        coerce = True
        strict = 'filter'
    
    index: int = pa.Field(ge=0)
    confirmation_date: pa.Date = pa.Field(nullable=True, coerce=True, default=pd.NaT)

data = [
    {'index': 1},
    {'index': 1}
]

# This raises SchemaError
validated_df = BaseValidationSchema.validate(pd.DataFrame(data))

Error Output

pandera.errors.SchemaError: expected series 'confirmation_date' to have type date:
failure cases:
   index failure_case
0      0          NaT
1      1          NaT

Expected behavior

The validation should pass successfully. When add_missing_columns = True is set and a column is missing from the input DataFrame, Pandera should add the column with the specified default value (pd.NaT) and recognize it as a valid null value for the pa.Date type since nullable=True is specified.

The expected output would be a DataFrame with both columns:

• children: int64 with values [1, 1]
• confirmation_date: datetime64[ns] (or appropriate date dtype) with NaT values

Desktop:

• OS: Ubuntu
• Browser: /
• Version: /

Additional context

I am aware of using datetime and then after converting to date but in my context this is not a viable option.