Add input validation and error handling to make debugging easier

[maxsch3]

When used in automated pipelines, it is very difficult to debug without validation of inputs and proper error handling and reporting. Need to make at least the following checks

1. Structure validation:

• Structure is not None,
• It is a list
• each element is a tuple

2. Data is a valid dataframe

[maxsch3]

BatchShaper.inverse_transform must check if y_structure is not None