Mixing feature: Precompute multi-resolution waveform peaks instead of redraws

[Stryker999]

Normal view, track playing: 1 style recalc per second, renderer at 2–3%, GPU process at ~0.5%.Editor view with just the  track playing does over 60 style re-calcs per second, renderer 40–55%, GPU 12–53%.(5070ti 64 gb ram i7 6700k if thats relevent)
Put the playhead on its own compositor layer , also decode 30 seconds of the audio into PCM decoded PCM already exists as a transient, 30 seconds of float PCM is ~10 MB, likely warrants a security review but holds decoded audio no longer than playback already does. As your well aware loopback capture of the audio output exists anyway — holding a scrub buffer changes nothing in that aspect.

Anyway i'm sure you already know all**bleep** on a much deeper level than i do and business is one prio to the next and this is a small niche part of a big picture.
Thanks.