feat(skills): add change complexity tier assessment to propose and new workflows

[marcoskichel]

Summary

• Adds a tier assessment step (step 2) to both /opsx:propose and /opsx:new workflows that fires right after the user describes their change, before any OpenSpec scaffold is created
• Classifies changes into three tiers and acts accordingly:
  • Tier A (trivial) — typo, formatting, comment-only, non-semantic refactor → warns the user a direct PR is sufficient, asks confirmation before proceeding
  • Tier B (small/low-risk) — minor UX/config tweak, refactor with preserved behavior, no API/data/security impact → suggests a lightweight schema, queries openspec schemas --json to find one
  • Tier C (meaningful) — behavior change, API/contract, migration, security/privacy, cross-team → proceeds silently with the full workflow

Motivation

OpenSpec's own contributing guidance recommends "progressive rigor" — small fixes go direct PR, larger changes use the full workflow. Without a gate, the agent defaults to full OpenSpec for everything, adding unnecessary overhead for trivial changes and eroding the value of the framework for the changes that actually matter.

Test plan

- [ ] Run `/opsx:propose` with a trivial change description (e.g. "fix typo in README") — expect Tier A warning + confirmation prompt
- [ ] Run `/opsx:propose` with a small change (e.g. "update button color") — expect Tier B lightweight suggestion
- [ ] Run `/opsx:propose` with a meaningful change (e.g. "add OAuth2 login flow") — expect no prompt, proceeds directly
- [ ] Same three scenarios for `/opsx:new`
- [ ] Build passes: `npm run build`