docs(challenge): experimental caveat, honesty review, and lives

[Larens94]

Adds clarity to the CodeDNA Challenge rules:

Experimental (agent × language)

• CodeDNA is still maturing; quality can vary by AI coding agent and language/framework.
• Participants should open issues or fix PRs when install/annotation/CLI misbehaves — that is part of the challenge.

Honesty & verification

• Metrics and narratives must be truthful (no invented results).
• Fabrizio / review team may request a live review (e.g. Google Meet) to walk through runs.
• Refusing a reasonable verification request can void prize eligibility.
• Submission templates include honesty checkboxes.

Lives

• When possible we will host public live sessions discussing submitted tests (optional unless asked for verification).

Touched: docs/challenge.md / .it.md, docs/challenge.html, submission templates, README EN/IT.