refactor(components): investigate touch/pointer events for mobile long-press reliability on `Tooltip`s

Description
Following our recent bugfix for rapid taps on mobile, we have uncovered a structural limitation in how we handle mobile long-presses for the Tooltip component, specifically on Android.

The current tooltip logic relies entirely on synthetic mouse/focus events (focusin, mouseenter). This creates a "focus trap" scenario:

1. Initial Interaction: A user taps a toggle button. The browser fires focusin, our tap-handler cancels the tooltip, and the button changes state. Crucially, the button retains focus.
2. Subsequent Long-Press: The user performs a long-press on the same button to read its tooltip.
3. The Failure: Because the button is already focused, the mobile browser does not dispatch a new focusin or mouseenter event. Our tooltip component remains completely unaware of the user's finger on the screen.
4. Workaround: The user must tap outside the button to remove focus (blur) before a long-press will work. This is unacceptable UX.

The Challenge
To fix this, we need to stop relying on legacy synthetic events for mobile and start listening to native touch interactions (touchstart/touchend or modern pointerdown/pointerup).

However, this is non-trivial. Implementing touch/pointer events requires:

• Carefully managing custom timers directly tied to the touch lifecycle.
• Preventing duplicate event firing (since mobile browsers will still fire mouseenter after a touchstart to support legacy code).
• Ensuring that scrolling the page doesn't accidentally trigger tooltips (handling touchmove or pointercancel).
• Guaranteeing that desktop interactions (hovering with a real mouse) remain completely unaffected and performant.

Next Steps (Brainstorming Required)
We need to discuss the architectural approach before writing code.

• Should we migrate the entire Tooltip event architecture to the unified PointerEvents API?
• Or should we keep the current mouse/keyboard events for desktop and surgically attach explicit touchstart/touchend listeners for mobile overrides?

Start by locating the Tooltip component and the recent rapid-tap mobile bugfix, then trace how focusin and mouseenter currently drive tooltip behavior. Compare a unified Pointer Events approach with explicit touch listeners, including timer ownership, duplicate events, cancellation during scrolling, and desktop behavior. Done means the project has an agreed architectural direction and clear acceptance criteria before implementation begins.