test: ten wiring tests name a condition they do not exercise ("bare in the single composition")

[morozsm]

Problem

Ten tests in the semantic-wiring suites are named after a claim they do not
test. They assert that a surface renders bare, and attribute that to the
composition:

it('renders it bare in the single composition, outside every zone', ...)

They pass — but not for the stated reason. These are standalone mounts that
resolve no surface plan: useSurfacePlan() falls back to NO_PLAN
(presentation/workspace/resolution.ts), so zoneOwning() returns null for
every surface regardless of what any manifest declares. Under desktop-v2 —
which has declared a zone for all 14 members of SEMANTIC_SURFACE_NAMES since
the rework tail — these same surfaces mount zoned, not bare.

So the name states a property of the composition; the test measures a property
of the fixture. A reader looking for "does the single composition zone this
surface?" is told the wrong answer by the name alone, without opening the file.

This is the test-name half of the comment defect swept in #2914 — that PR
corrected the prose and deliberately left test names alone, since renaming is a
code change.

The house pattern already exists

semantic-tx-aux-wiring.component.test.ts:381 is already named correctly:

it('binds no zone id to the txAux surface in either composition, absent a resolved plan', ...)

and its comment states the rule the others elide: "a zone element must exist
only where BOTH a layout declared one AND a plan actually resolved it." The
same file pairs it with a positive counterpart at :788 that binds the real zone
under resolveSurfacePlan(desktopV2Layout, ...). That pair is the model.

Sites

frontend/src/components-v2/wiring/__tests__/ — renders it bare in the single composition, outside every zone:

• semantic-antenna-wiring.component.test.ts:243
• semantic-band-wiring.component.test.ts:209
• semantic-cw-keyer-wiring.component.test.ts:653
• semantic-dsp-wiring.component.test.ts:319
• semantic-rf-front-end-wiring.component.test.ts:362
• semantic-ritxit-scan-wiring.component.test.ts:416
• semantic-rx-audio-wiring.component.test.ts:409
• semantic-scope-controls-wiring.component.test.ts:296

Same shape, binds no zone id ... in either composition:

• semantic-meters-wiring.component.test.ts:313
• semantic-scope-display-wiring.component.test.ts:309

Already correct, do not touch: semantic-tx-aux-wiring.component.test.ts:381.
Out of scope: ScopeDisplaySurface.test.ts:121 and MetersSurface.test.ts:211
are pure-surface tests — "binds no zone id of its own" is a true statement
about the surface, not about a composition.

Acceptance criteria

1. Each of the ten names states the condition actually exercised — that no
   surface plan is resolved — rather than attributing bareness to the
   composition. Follow semantic-tx-aux-wiring:381's wording.
2. No assertion body changes. This is a naming fix; if a body needs to change,
   that is a different defect and belongs in its own issue.
3. Whether each suite also gains a positive counterpart (the surface IS zoned
   under resolveSurfacePlan(desktopV2Layout, ...), as tx-aux:788 does) is a
   judgement call per suite — decide explicitly and record the decision, rather
   than adding ten by reflex. The vocabulary-wide guarantee is already held by
   presentation/layouts/__tests__/zone-ownership-coverage.test.ts.
4. Full suite green: cd frontend && npx vitest run.

Why it matters

zone-ownership-coverage.test.ts records that the MOR-1317 ledger is empty —
every surface is zone-owned on desktop-v2. Ten test names in the suites that
exercise those very surfaces still say the opposite. The next implementer reads
the name, not the fixture.