New mixed embedding

[lea-33]

Generate a new, more structures mixed embedding.
The text for the mixed embedding should have a clear, dictionary like structure with the following entries:

• Content: 1 sentence about the slides content
• Style: what kind of slide it is - options: Opening/Greeting, Ending/Closing, References, Divider/New Topic Introduction, Text only, Image only, Mixed Content
• Language: options: English, German, Mixed
• Knowledge Level: what's the target group for this slide - options: beginner, intermediate, expert, none
• Learning perspective: what's the main learning objective for this slide - options: Defining concepts, Explaining concepts, Comparing concepts, Visualizing Concepts, Engage the Reader, none

This list is just a first idea and can be adjusted.
By using this structure for the mixed embedding, we can get a more meaningful structured embedding space. Also we could save the output itself and use it later on for classification or other use cases.