format: hold the text and picture families to the declaration they share

[donislawdev]

textenc.Properties() becomes textenc.Axes(names...) alongside textenc.Names(), the shape the archive package has had since the containers were unified. All three text formats take both axes and call it with both, so the narrowing has no dead call. It is there because the fourth format is already named and cannot take both: an HTML document has to be UTF-8 by its own specification, so the only encoding axis it could carry is the mark.

The gap was measured, not suspected

Two drifts were introduced on the tree, run, and reverted.

what was broken	guards that stayed green	what a user saw
md declares encoding itself - same kind, unit and default, one encoding fewer	11	md goes on writing utf-16be and stops offering it, so txt and xml answer one question and md answers another
bmp declares width itself, Min: 0	10	tfg formats bmp prints from 0 to 20000 while --set width=0 is refused with between 1 and 20000

The control matters: moving the default turned TestADeclaredDefaultIsTheOneTheFormatUses red at once. That half was already closed, which is why the mutation in the suite keeps the default in agreement - otherwise it would be proving another guard.

The second drift was caught only by the backstop in imagedim.Value, the one picturesides_test.go says a person cannot reach. A person reaches it exactly when a declaration drifts, and that was the case with no guard.

Why the picture guard is not a straight comparison

Largest, Default and Detail belong to the format and are supposed to differ - five ceilings, one default, and four sentences of which three would be made wrong by a shared one. So the guard rebuilds the declaration from the parts a format supplies and requires the result to be what is registered. That covers a field nobody has thought of yet, which a list of field comparisons would not.

Narrowing by name, not by value

No text format wants a narrower set of encodings, and all ten picture formats call Width and Height in pairs. Narrowing a value is what imagedim.Side already does, and it was built after ten different ceilings had been measured rather than before.

Evidence

• tfg formats --json hashes to 0b1361a7294ab5b71b6acd39f13c9c74effa0a6df215f241fc3989636836dfc7 before and after
• full suite green, ok internal/guard 275.8s
• journey.py 300 of 300
• both new mutations caught, staleness.py 919 patterns each occurring once
• gofmt (from GOROOT) and go vet clean

Known limit, written down rather than fixed quietly

Both guards skip a name listed in deliberateHomonyms, and they skip it for every format rather than only for the one it excuses. Writing encoding there would silence half of the first guard. archiveaxes_test.go has no such door because it keeps no list. Narrowing the skip means changing the shape of another guard's list, so it is recorded in the guard's own comment and in the document.

🤖 Generated with Claude Code