workflow templates

Fix “works on my machine” for ML: this article delivers a paste-and-run, Bash-first AI workflow (Makefile + .env + scripts/src) that runs on any Linux with Python/make, standardizes data→train→evaluate, uses env vars and idempotent steps, logs and packages artifacts, includes a stdlib Naive Bayes demo, and scales later to cron/systemd, containers, cloud, or GPUs without changing the shell contract.