Title: Fix Scoping Issue and Enhance Robustness in Image Loading Functions

Description:
The current implementation of the load_dataset and load_images_and_labels functions has several issues and can be improved for better robustness and clarity. This issue aims to address the following problems:

Scoping Issue: The label_dict variable is not accessible in the load_dataset function because it is defined within load_images_and_labels and not returned.

Error Handling: There is no error handling for image loading, which can cause the entire process to fail if a single image is problematic.
File Extension Check: The file extension check for images is case-sensitive, which might cause valid image files to be skipped if their extensions are in uppercase.
Code Clarity: Improving the readability and structure of the code by adding missing types in docstrings and making minor enhancements.

Proposed Fix:

Return label_dict: Modify load_images_and_labels to return label_dict along with X and y, and adjust load_dataset to accept and use this return value.
Add Error Handling: Introduce error handling for image loading to skip files that cause exceptions, ensuring that one problematic file does not halt the process.
Case-Insensitive Extension Check: Use str.lower() to make the file extension check case-insensitive, ensuring all valid image files are processed regardless of their extension case.
Code Enhancements: Update docstrings to include missing types and enhance the overall readability of the code.