sage-pond/Punkttokenizer — reverse-engineered prompt

Build me a simple, lightweight Luganda sentence splitter from this repo. I want it to use the included trained Punkt model in the JSON file and turn a block of Luganda text into a clean list of sentences.

Please make it easy to use for someone who just wants to paste text and get sentence breaks back. A small Python script is enough, but if it makes sense, also add a tiny command line option where I can pass in a text file or type text directly. Include a clear example using Luganda text so I can see it working right away.

Keep it focused on sentence boundary detection, not translation or anything bigger. Since this model is meant to handle abbreviations and dense text better than naive splitting, make sure the setup loads the model correctly and uses it the right way. Add short setup instructions and a quick note on how to run it. If anything in the README is outdated or unclear, look up the current docs online if you need to.