Add annotated guide to NanoTabPFN architecture

[StatMixedML]

Description

This PR adds "The Annotated NanoTabPFN" - a detailed resource that explains the model architecture and implementation details, inspired by "The Annotated Transformer".

What this adds

• Complete walkthrough of the NanoTabPFN model architecture
• Step-by-step breakdown of each component (FeatureEncoder, TargetEncoder, TransformerStack, etc.)

Why this is valuable

• Makes the codebase more accessible to students and researchers
• Serves as a learning resource for understanding TabPFN basics

The guide is designed to complement the existing code, making it easier for others to understand, modify, and build upon NanoTabPFN.

Would love feedback on clarity, technical accuracy, and any additional topics that should be covered!

[StatMixedML]

This should be the final version.

@JohannesHog & @AlexanderPfefferle Happy to receive your comments.

[AlexanderPfefferle]

I will try to take a look at it once I find the time.
Thank you!