Thank you Mario — this is extremely helpful feedback on both fronts.
Noted on the format: post as trailer, not film. I’ll follow that structure for any future submissions.
And thank you for going through each finding. The context on the three-pass problem is particularly clarifying — knowing that (a) and (b) are effectively cheap in practice changes the priority picture considerably. The clearGlobalCache chicken-and-egg observation is fascinating; it sounds like the cache can’t be trusted precisely because it gets cleared so aggressively, which means fixing the clearing strategy and building trust in the cache would need to happen together.