Spec question: may a frozen principal close itself?

A question for the spec author, surfaced during the server production-readiness effort (PR #23's roadmap) and so far tracked only as prose.

SPEC.md does not state whether a principal in the frozen state may issue its own close. The server needs a deterministic answer: when a close commit arrives for a frozen principal, signed by that principal's own active key, must the server accept it as a valid lifecycle transition or reject it as a mutation of frozen state?

Arguments exist in both directions — freezing exists to stop chain mutation, but close is the one terminal transition a principal may plausibly retain the right to make about itself. The spec should say which reading is normative.

Resolution belongs in SPEC.md; implementations follow whatever it decides.

Read SPEC.md alongside the frozen-state and close-transition language, and review the server production-readiness context in PR #23's roadmap. Resolve whether a frozen principal may close itself, then record the chosen interpretation as a normative rule in SPEC.md so implementations have a deterministic answer.

Written by the indexing model from the issue text.