feat: support polars with auto detection from TaskOnKart's Generic type

[kitagry]

Summary

This PR adds support for polars DataFrames alongside pandas DataFrames in gokart's file processors. Ta sks can now specify their DataFrame type via the TaskOnKart[T] type parameter, and the appropriate pro cessor will be automatically selected.

Features

• Multi-backend support: File processors now support both pandas and polars DataFrames
  • CsvFileProcessor
  • JsonFileProcessor
  • ParquetFileProcessor
  • FeatherFileProcessor
• Auto-detection: Automatically detects DataFrame type from TaskOnKart[pd.DataFrame] or TaskOnK art[pl.DataFrame] type parameters
• Backward compatible: Defaults to pandas when no type parameter is specified

Usage Example

import polars as pl
from gokart import TaskOnKart

class MyPolarsTask(TaskOnKart[pl.DataFrame]):
    def output(self):
        return self.make_target('path/to/target.feather')

    def run(self):
        df = pl.DataFrame({'a': [1, 2, 3]})
        self.dump(df)  # Automatically uses polars-compatible processor

class MyPandasTask(TaskOnKart[pd.DataFrame]):
    def output(self):
        return self.make_target('path/to/target.feather')

    def run(self):
        df = pd.DataFrame({'a': [1, 2, 3]})
        self.dump(df)  # Uses pandas processor (default behavior)

Why not #457 ?

In #457, we switch between Polars and Pandas using GOKART_DATAFRAME_FRAMEWORK. However, this implies that we cannot use both at the same time. Since projects often migrate from Pandas gradually, we should allow users to use both Polars and Pandas simultaneously.