Image checksumming and visual regression

These helpers exist for autotesting. They compare or persist perceptual hashes of the rendered scene.

Function	Returns	Description
`GetVisibleInputChecksumString(ignoreLivePreview)`	`string`	Computes an MD5 checksum (as upper-case hex) over the concatenated GUIDs of all inputs that are currently being rendered. When `ignoreLivePreview` is true, internal live-preview scenes are excluded from the set. Returns an empty string if there are no visible inputs.
`CompareImageReference()`	`float`	Compares the current rendered frame against the reference image for this project. Returns the percentage of differing pixels (0-100), or -1 if no reference image exists or an error occurred. Designed for polling from OnRenderFrame in test scripts.
`GetImageChecksum()`	`string`	Computes a perceptual hash (checksum) of the current rendered frame from the topmost scene. Returns a 16-character hex string representing a 64-bit hash that is tolerant of minor pixel-level differences — visually similar frames produce the same hash. Returns an empty string if no frame is available or an error occurs.
`GetImageChecksum(sceneName)`	`string`	Computes a perceptual hash (checksum) of the current rendered frame from the specified scene. Returns a 16-character hex string representing a 64-bit hash that is tolerant of minor pixel-level differences — visually similar frames produce the same hash. Returns an empty string if no frame is available or an error occurs.
`CompareOrStoreImageChecksum(sceneName, stepName)`	`bool`	Compares or stores a perceptual image checksum for a named step. The checksums are persisted in a JSON file alongside the project file, named -step-img-checksum.json. The file contains a flat object mapping step names to checksum hex strings. - If the step does not exist in the file, the current checksum is stored and the method returns true. - If the step already exists, the current checksum is compared with the stored value. Returns true on match, false on mismatch.

function OnRenderFrame() {
    if (frameCount == 30 && !Project.CompareOrStoreImageChecksum(null, "after-intro"))
        Logger.Error("Visual regression at after-intro");
}