Chunking vs Compaction for reading the issue

Options for breaking down the issue to fit into context: 
- produce a summary (encoded?)
- persist in external memory (files/db)
- delete irrelevant tokens
- store in vectorDB