feat: add Agent Resilience and Fuzzing skill

[LorenzaVolponi]

🚀 Summary
This PR adds a new skill agent-resilience-and-fuzzing that guides agents through resilience and fuzzing testing for any skill, plus a helper script (fuzz-skill.sh) to automate simple fuzz tests. 🛡️

💡 Motivation
While agent-skills already covers many quality dimensions (TDD, code review, security, etc.), there is no explicit workflow for ensuring that skills themselves are resilient to unexpected or adversarial inputs. 📉

As agents get more autonomous, a malformed input or an unexpected API response can cascade into a broken state. This skill:

🛡️ Defines a concrete process for identifying and testing edge cases in any skill.
🤖 Ships a generic fuzzing helper script that can be reused across the repo.
🧩 Integrates with the existing skill anatomy by adding fuzz-cases.md, RESILIENCE.md, and resilience-related verification steps.

🧪 How to test

Run the included script against an existing skill:
./skills/agent-resilience-and-fuzzing/scripts/fuzz-skill.sh \ --skill skills/incremental-implementation/SKILL.md \ --arg "limit" \ --type number \ --fuzz-type boundary,mismatch,overflow

Verify that the script generates and runs fuzz cases, clearly showing which cases are OK vs FAILED. ✅

✅ Checklist

New skill under skills/ with kebab-case name
SKILL.md follows docs/skill-anatomy.md
YAML frontmatter with valid name and description
Description starts with what the skill does, then Use when triggers
Skill is specific, verifiable, and minimal
No duplicate content; other skills are referenced by name
No empty scripts/ directory; this skill ships a concrete, usable script

[nucliweb]

A SKILL cannot be placed within a Markdown block