Feature/support for loading datasets from dlt within databricks

[chakravarthik27]

This pull request includes several changes to the langtest/datahandler/datasource.py and langtest/tasks/task.py files to add support for Spark datasets and improve the handling of file extensions. The most important changes include the addition of a new SparkDataset class and modifications to the __init__ and load methods to accommodate the new dataset type.

Support for Spark datasets:

• Added a new SparkDataset class to handle Spark datasets, including methods for loading raw data and preprocessed data, and initializing a Spark session. (langtest/datahandler/datasource.py)

Improvements to file extension handling:

• Updated the __init__ method to set file_ext from the file_path dictionary if provided, and adjusted the logic to handle cases where the source key is present. (langtest/datahandler/datasource.py) [1] [2]
• Modified the load method to include "spark" as a valid file extension for initializing the data source class. (langtest/datahandler/datasource.py)

Enhancements to task handling:

• Updated the create_sample method to ensure labels are converted to strings before being added to the list. (langtest/tasks/task.py)