Enable Off-Screen Coordinate Pixel Calculation (Bypass .fit() Transformation)

## New Feature

I propose an enhancement to the iOS SDK’s coordinate-to-screen-point conversion. Currently, the `point(for:)` function returns (-1, -1) for off-screen coordinates due to the final `.fit(to: size)` call:

```swift
public func point(for coordinate: CLLocationCoordinate2D) -> CGPoint {
return __map.pixelForCoordinate(for: coordinate).point.fit(to: size)
}
```
In my use case —with a constant map pitch of 0 and coordinates close to the screen— I need access to the raw pixel values even if the point is off-screen. I suggest either:

* Exposing map.pixelForCoordinate directly, or
* Adding an alternative function (e.g., unsafePixelForCoordinate) that omits the .fit() transformation, along with clear documentation warning about its potential pitfalls.

## Why
This feature is valuable for developing custom overlays, annotations, and animations that require off-screen coordinate calculations. While Mapbox Kotlin SDK developers might use reflection to access hidden methods, this approach is neither desirable nor possible in Swift.

Start with the iOS SDK's point(for:) function and the referenced __map.pixelForCoordinate call. Trace how the final .fit(to: size) transformation handles off-screen points, then review the public API and documentation conventions. Done means an agreed, documented way to obtain raw off-screen pixel values without changing existing point(for:) behavior.

Written by the indexing model from the issue text.