An extreme value needs a second, differently-shaped method

[h4x0r]

§5.3:

A universal negative or a 100% result MUST be re-derived by a second,
differently-shaped method before reporting. Populations rarely agree perfectly; such
a result more often measures the instrument than the world.

peira already enforces the neighbouring rule — §5.2, "universal quantifiers must be
scoped to what was actually queried" — as PEIR-CLASS-EXTENSION-UNDECLARED (白馬非馬),
which blocks a claim that quantifies universal without declaring its extension.

What is missing is the evidential requirement rather than the scoping one: having
declared the extension honestly, a universal negative still needs two differently
shaped methods behind it. An extreme value is exactly where a single method is least
trustworthy.

Proposed PEIR-EXTREME-SINGLE-METHOD — a claim whose quantifier is universal and
whose polarity is negative, supported by observations from a single instrument.

Depends on the Instrument node kind; "differently-shaped method" is not computable
while every observation is anonymous as to its source.

Source: ~/.claude/knowledge/evidence-claim-grading-spec.md §5.3, and §5.2 for the
part already enforced.