The attached files are entirely synthetic. Write a small, reproducible analysis
for a real CSV with the same columns.

Requirements:
- accept --input and --output command-line arguments;
- never contact a network service;
- never modify the input file;
- validate required columns and explain errors without printing row data;
- save the analysis code, a summary CSV, one plot, and a short run log;
- list required Python packages with versions;
- do not hard-code values from the synthetic rows;
- include a command that tests the code on example.csv.

Explain the plan first. Then provide the files and exact test command.