Well if I was a developer I would definitely be toying around with some lag-reducing ideas(which I understand is already a priority, but hey OP asked me what Id do). I can't help but wonder if a good chunk of the lag is from event trigger overhead, since I noticed a sharp reduction in lag after beating the PULSE Tangrowth. I'd try making the event cycle through a longer animation sequence, so that while each animation event would be larger, theyd be triggered far far less. The question of that working or not would be if reducing the overhead saves more than making longer events creates. Id place my money on yes, but no way to tell without trying it, but itd be a pain to go through all those events like in the wasteland, only to potentially have it not be worth it. Also by default RMXP uses an old version of the RGSS which has long had known performance issues, and by extension pokemon essentials uses it as well. However, I have seen an interesting guide on importing the updated engine from RMVX, but the guide wasn't clear on how it may break scripts so while its pretty easy to do theres the chance that itd require massive effort to make pokemon essentials to be compatible and/or itd have to be a separate release makred as an essentially perma-unstable release. I suppose this is something I could actually play with my own copy of pokemon essentials, but even if I prove itd be possible theres no way I'd be able to bug test everything that could go wrong. Still a five times performance boost to the core engine would be nifty.
I second the notion of having a fairly early move tutor that only teaches moves to undervalued pokemon. This would also be a very subtle way to reward players for picking less common starters without appearing to punish the player for choosing an OP starter that they liked. Simply pokemon like blaziken wouldnt be able to learn anything useful, while your patrats and chikoritas of the world can get some nifty tricks to use. The exact specifics of what moves, what pokemon, and when/where to place such a tutor would take a bit of fine tuning though.
The magikarp idea would be an amusing joke, even though I started playing after it got removed. Easy fix would be to just make it genderless. Since you could make it as a separate mon, by default youd just have it not able to evolve into gyrados citing in game something about pollution inhibiting its potential. People get their joke pokemon and eat it too.
Speaking of eating pokemon, I think more underused pokemon should have small chances of appearing with useful(but not overpowered items). A good example for me is my most memorable moment in Reborn. At first I was put off by the removal of a lot of strong pokemon cause it felt like artificial difficulty. Then I found trubbish. That little gem came with a rare black sludge and has been a cornerstone in my team well past his real usefulness, simply because I became attached to it. I felt rewarded for going with this little trash pile I wouldnt have even bothered catching in the main games. It was never overwhelmingly strong, just like I said it felt rewarding seeing the tricks it managed to pull off thanks to that bit of recovery every turn.
As a quick aside, that giratina video someone posted was really stupid. Most of it has just been fan theories, nothing more and occasionally far less. I had to stop watching when he seriously tried to claim the lavender town creepy-pasta was true(yes the song got changed cause kids got headaches and shit, not there wasn't weird mass suicides and investigators dieing mysteriously).
edit: Tried my hand of importing the engine into raw essentials, unfortunately essentials is so heavily modified, that the guide for doing so doesn't really apply. Fixing the syntax conversions from ruby 1.8 to 1.9 is easy, but when I get unhelpful error messages like 'the specified procedure could not be found' not much I can do(maybe if it specified the specified procedure Id have something to work with).
2nd edit: Fixed it, reading is op. Time for more syntax changing!