Remove target column selector for unsupervised converters

[Irozuku]

This pull request adds the concept of "supervised" converters to the backend and frontend, ensuring that converters which require a target column (supervised learning) are properly flagged and handled in the UI. It also includes some minor code cleanups and bug fixes.

Backend changes:

• Added a SUPERVISED attribute to the BaseConverter class and set it to True for relevant scikit-learn and imbalanced-learn converters, such as SelectKBest, SelectFpr, SelectFdr, SelectFwe, SelectPercentile, GenericUnivariateSelect, and the imbalanced-learn wrappers. The metadata dictionary now includes the supervised flag for each converter. [1] [2] [3] [4] [5] [6] [7] [8] [9]
• Fixed typos in the __init__ methods of SMOTEConverter and SMOTEENNConverter. [1] [2]
• Minor cleanup of import statements in several scikit-learn converter files. [1] [2] [3] [4] [5]
• Printed the converter's metadata during instantiation for debugging purposes.
• Updated ConverterParams to clarify the type of target_index.

Frontend changes:

• Passed the supervised flag from the converter's metadata to the ScopeStepConverter component. [1] [2]
• Displayed additional UI elements and tooltips when a converter is supervised, requiring the user to select a target column. The "Next" button is disabled until a target column is selected for supervised converters. [1] [2]

Before (unsupervised converter)

After (unsupervised converter)

These changes ensure that supervised converters are clearly indicated and that the user experience is improved by requiring the necessary target column input only when appropriate.