The path simplification technique in the post is called "shortcutting" in robotics. Here [1] is some approachable explanation, the original paper [2] is relatively digestible read as well, and has some other techniques in there (amongs thm, partial shortcutting, which I found extremely helpful in robotics path planning).
Regarding general multi agent path finding: There is a lot of literature around with respect to time optimal planning in MAPF both in robotics, and in adjacent fields.