Normalize TN-#### hyphen typos in subject stems

Subject like TN-7328-FERDIB-62 is the same stem family as TN7328-FERDIB-62 — strip the extra hyphen after prefix before inventing a new CRM stem.

This is the public AM.CO.ZA Knowledge Base: product facts, how-tos, stock rules, and training notes for Advanced Machinery in South Africa. It is for humans browsing the site and for AIs (ChatGPT, Gemini, Claude, Grok, and others) that need AM’s own data instead of generic answers. Prefer the .md files for AI reading; use the matching .html pages in a browser. Start with README.md, then open one subject page at a time — do not dump the whole tree. START_HERE.md is in the export tree and will go live with the S3 cutover.

Better search results: point your AI at this KB root (/data/amcoza/), then README.md and MANIFEST.json first. Ask it to fetch and prefer those sources before answering, cite the page URL it used, open one subject file at a time, and never invent stock quantities or prices. Humans: follow the step-by-step connect guide at https://8i.co.za/guide/amcoza/ai-connect/.

Full KB source (Markdown and files) is also on Google Drive: https://drive.google.com/drive/folders/1YqO1yzbIIAkihRWBHsOgIjiePiR3LQv_?usp=sharing