Construct this against an output stream.
The stream to render to (default stderr).
Minimum ms between live lines when not a TTY.
Finish rendering: freeze any running stage's line and print the separate before/after Total line.
OptionalatElapsedMs: numberOptional load-end elapsed ms, so the final stage's duration covers the time until the load finished (not just until its last progress event).
OptionalatMemoryMb: numberOptional load-end heap MB.
Print the early model line (from header info).
The model header info.
ReadonlyonProgress callback — bound as an arrow so it can be handed directly to a ProgressTracker / ModelLoadOptions.onProgress.
The progress event to render.
Renders load progress on a terminal using the shared normalized load-log format (core/progress_log.ts) — the same text rendition Share mirrors to the browser console, so a pasted CLI run and a pasted browser log read identically. Format spec: Share design/new/load-log-format.md.
Writes to stderr so stdout stays clean for data output. On a TTY the current stage's line repaints in place and freezes (newline) when the stage completes; when redirected (CI logs) it prints the live line at most once per nonTtyIntervalMs plus every frozen stage line.