fix(announcements): float the banner instead of occupying layout

[philmerrell]

Follow-up to #976. Dismissing the banner pulled the whole view up by its height, because it was a flex child of the shell's <main> — appearing and disappearing reflowed everything below it. Reserving the space forever would have been a worse fix.

It is now positioned absolute against a relative <main>: a rounded, shadowed pill floating over the content rather than a full-bleed strip displacing it.

Measured before and after a dismissal, every content element moves by exactly 0px in both axes:

	Δtop	Δleft	Δheight
#app-scroll-container	0	0	0
greeting	0	0	0
composer	0	0	0

It deletes more than it adds

Once the banner stops occupying space, nothing needs to know its size — so the whole measurement mechanism and all three offsets from #976 come out:

• --announcement-banner-height, its ResizeObserver, the height signal, and the DOCUMENT / ElementRef / DestroyRef injections
• .chat-topnav-wrapper → back to top: 0
• .chat-container-empty.full-page → back to inset: 0
• both floating sidenav control clusters → back to top-4

Net −111/+83 lines.

Two details worth reviewing

top-16 is the single constant that replaces all of it, and it is not arbitrary. On a chat route it lands the pill immediately below the fixed topnav — the placement §D1 asks for — and everywhere else it clears the shell's floating sidebar buttons at top-4, which a centred pill would otherwise overlap on any viewport narrow enough for the two to meet. Verified at 375px: controls end at y=56, pill starts at y=64, overlapsControls: false. One constant, no route awareness.

The positioning strip spans the full content width, so it is pointer-events-none with pointer-events-auto on the pill alone. Without that, an invisible band would swallow every click aimed at the topnav and the sidebar buttons underneath it. Verified: a click 30px outside the pill resolves to .chat-container-empty, not the banner.

relative on <main> is load-bearing — without it the pill anchors to the viewport and drifts out from under the sidenav's padding transition.

Verification

Browser-verified against real dev data with a local app-api: light and dark, desktop and 375px, no horizontal overflow, hostLeft: 288 confirming it anchors to the content region rather than the viewport, and the dismiss still writing its durable server-side ack.

Specs updated in place — the two height-publishing tests are replaced by five overlay ones covering absolute positioning, click pass-through, the top-16 clearance, the rounded-card styling, and the absence of any document-level layout variable. Full frontend suite 2474 passed.

Test data created in dev was cleaned up; only real users' acks and the existing "Welcome to Dev" announcement remain.

Independent of #978 (PR-6) — no shared files, checked rather than assumed.

🤖 Generated with Claude Code