Fix /setup environment selection (render Markdown table instead of scraping terminal)

[amilandi]

Problem

Running /setup and choosing "Yes, list my environments" frequently shows no environment table (or a truncated one), and Copilot Chat often doesn't know how to proceed.

Root cause: step1.md ran discover.py --list-environments (which prints a plain-text ASCII table to the terminal) and then asked the model to read that terminal output, parse every row, and rebuild each environment into a vscode_askQuestions picker. Copilot Chat truncates terminal scrollback and never surfaces the ASCII table to the user, so the reconstruction lost rows or failed entirely.

Fix

Stop relying on the model scraping the terminal. discover.py --list-environments now also writes a complete, numbered Markdown table to workspace/onboarding/environments.md and prints an ENV_TABLE_MARKDOWN:<path> marker. step1.md reads that file, shows the full table to the user verbatim, asks them to type the row number, then runs --select N.

Benefits: no terminal scraping, no picker reconstruction, the user always sees the full table, and it scales to any number of environments.

Changes

• list_environments.py: new write_environment_markdown() (pipe-escaped names, (no Dataverse linked) placeholder).
• discover.py: write the Markdown file and emit the ENV_TABLE_MARKDOWN: marker on the list (non-select) path.
• src/skills/onboarding/step1.md: §1.1/§1.1a/§1.1b now render the table from the file and use number-entry selection with an invalid-number retry.
• tests/scripts/test_discover.py: cover the Markdown writer (numbering, placeholder, pipe escaping).

Validation

• New + existing tests/scripts/test_discover.py pass (16).
• Pre-existing failures in test_installer_maker_profile.py are unrelated (they also fail on main).

[nehaoss]

👋 Friendly reminder: This PR has been inactive for a while. Could the author or reviewers take a look and either push it forward, request changes, or close it if it's no longer needed? Let's keep our PR queue healthy! 🚀