Enhancement: Automatically compute and display token count on document upload, with context-aware warnings and suggestions

[flefevre]

Description:
When users upload a document to Open WebUI, especially for use in knowledge bases, it would be extremely helpful to:

1. Automatically compute and display the number of tokens the document will generate upon upload.

2. Warn users if the document exceeds the current model's context window, based on detected token count and the known token limit (e.g., 2048 for some local models).

3. Provide clear options:

• Create a new knowledge base for better content isolation.
• Add the document to an existing knowledge base if context space allows.

This enhancement would improve transparency and user control over context usage, aligning with the current limitations described in the documentation.

Rationale:
While Open WebUI offers options like "Bypass Embedding and Retrieval" or "Full Context Mode", users may not always understand when their document exceeds the context window, leading to truncated content or ineffective results. A proactive token count feature with contextual suggestions would bridge that gap and enhance usability.
https://openwebui.com/troubleshooting/rag/#2-only-a-small-part-of-the-document-is-being-used-%EF%B8%8F%EF%B8%8F

Possible UX Implementation:

• Show token count in the upload modal or post-upload.
• Display a warning badge or tooltip if the token count exceeds typical context window sizes.

Add buttons like:

• "Create new knowledge base with this document"
• "Add to existing knowledge base"