FEAT-09: --split <max-tokens> to chunk output across multiple files

[subhan-f]

Add a --split <max-tokens> flag that splits the output into multiple files (context-1.json, context-2.json, etc.) each staying under the specified token budget. Essential for large codebases that exceed a model's context window.

excod . --split 100000 --output context.json
# Produces context-1.json, context-2.json, context-3.json

Priority: Medium