v0.5.16

Added

• ManimWeb: run a manim-web (browser Manim) scene inline in a notebook. The engine is loaded from a CDN and your scene JavaScript runs with the manim namespace, a container element, and width/height/model in scope. Use manim-web's own Player for a full playback UI (play/pause, scrub timeline with segment markers, speed, fullscreen, export; autoPlay/loop/backgroundColor options). The scene source (code or src) can be inline JavaScript, a local .js file path, or an http(s):// URL — files and URLs are resolved in Python at construction time, so the browser always gets plain JS. JS runtime errors surface via the error traitlet.
• AsyncFlow: a new widget that traces a single async run live and renders it as a swimlane timeline. await AsyncFlow.trace(main()) runs the coroutine on the notebook's own event loop and streams task activity into the widget as it happens — one lane per task, solid bars for running and hatched bars for suspended-at-await, with child tasks nested under their parent (depth-first) and per-task timing (ran vs waited) on hover. Capture uses asyncio task hooks plus sys.monitoring, so it requires Python 3.12+.
• FramePlayer: play a sequence of images as an inline, optionally-looping "video". Accepts PIL images, file paths, URLs, bytes, base64 strings, or matplotlib figures (mixed is fine), and renders the current frame with play/pause/loop controls and a scrubber — no second cell that reads a slider value and re-renders.