Suicide levers were a pretty bad idea--a little thinking and you can come up with a solution that, you know, actually stops the bike. Another worst "invention": ceramic bearings. These are expensive and their properties have very limited advantage when applied to bicycles, aside from weight. If low friction or service life are the goal, high end stainless bearings typically outperform ceramic at a fraction of the cost.
I wouldn't put plastic on the list. There are a great many places where plastic is a suitable and appropriate material on bicycles--lever bodies, some derailleur body parts, saddles, bottle cages, etc. I also don't love airless tires, and what you write is essentially correct, but they really have been getting better and I'll admit there are use cases where they're not an outright terrible idea.
Really your best list mostly can be satisfied by bicycle designs from the early 1930s. Really at that point the best bicycles were very definitely useful, utilitarian vehicles that could travel in most terrain that bikes can now at not wildly dissimilar speeds, and most improvements since then have honestly been fairly evolutionary. The largest developments are in the development of the mountain bike, which really opens up a fundamentally different kind of riding on terrain that would have been impassible before. Not to mention how many developments for road bikes were predominantly developed to address the needs of mountain bikes (freehubs, threadless headsets, tubeless, 1x drivetrains).
Might be fun to compile of best bicycle innovations in, say, a narrower window of time. In relatively recent years, dropper seatposts, tubeless, 1x drivetrains, clutched derailleurs, hydraulic disc brakes have offered real world advantages to riders, for example, though they would never make a top 10 list for all time.
A lot of my worst headaches as a working mechanic aren't really terrible ideas but awful implementation. Press fit assemblies can exceed the theoretical maximum tolerances of threaded assemblies, but too many manufacturers have awful tolerances and poor implementations. Internal routing is of arguable benefit in some cases, and does look clean, but too many implementations are needlessly nightmarish, and I'm often a little offended when its used on utilitarian or entry level bikes because it makes amateur repair harder to get into. I find it ridiculous when cable actuated disc brakes end up working significantly worse than linear pull brakes on a Walmart bike.