Is it going to make too little sense to add angled subscale engines in the interstage area in quadcopter configuration, in a manner that resembles Porsche 918? That'll look cool, and could mitigate propellant feed problem too.
also, combustion devices are always one more point of complex and catastrophic mihaps. better keep it as simple as possible and don't miss.
There's already fuel line going into the fins as a ghetto hydraulics, so it should be more of "just" adding the oxidizer line rather than completely redoing that area for something that hasn't been done. Plus it could make the entire stack more flush and sleek.
The other problem is fuel: while the stage would be getting lighter burning fuel the whole way down, it's also slowing down so you spend more time in the descent stage, which in turn is increasing your engine durability and cooling requirements and increasing your fuel use etc.
And after all that, you're either (1) coming to zero velocity on the descent engines alone or (2) still need the Raptors to relight so you can actually land.
In the case of (1) you're hefting around a bunch of engines you're not using most of the flight, and expending more fuel etc. on the take off and landing. In the case of (2) well, why not just get the Raptors to be super-reliable and do the braking burn as-is?
As it is, we've now seen the Superheavy can do a landing burn with an engine failure to boot - improving from there takes us from those initial Falcon 9 landings to where are today (so common it's not news).
The current Raptor setup seems to be working pretty well for boostback and entry burns while the vehicle is in total vacuum, but it seem to get sketchier as propellants are expended and TWR increases, while altitude coincidentally reduces and aerodynamic stress increases. Almost every materials are stronger against tensile stresses than compressive stresses, and stainless steel is no exception, so an array of mighty Raptors trying to compress the booster lengthwise from the bottom, could be, just could be, less ideal than a pair of tip-jets pulling upwards and straightening the shell.
As for them being dead weight on ascent, I don't know, but as I understand it, every bit of thrust buys a bit of payloads at very moment of launch, then thrust requirement rapidly decreases and effect of Isp becomes more dominant. So I think there could be potentially some net positive effect in "test firing" the proposed side-jets at launch for few moments, depending on various parameters. I'm way too much of a layperson on actual rocketry to be even remotely sure, though.
I mean, picture Superheavy coming in to launch site, then the side facing retro-rockets light up as high as 100m above the ground, suspending the booster under it and sliding it gently onto the chopsticks. The retro-rockets are slightly angled, like Dragon 2 thrusters or slightly more angled. Isn't that going to look even cooler than what it is already?
Maybe it's less ideal, but SpaceX has almost certainly run the simulations to figure out whether the booster can indeed survive a late-stage launch configuration, and they probably found that the current design is perfectly capable of withstanding the stresses.
> So I think there could be potentially some net positive effect in "test firing" the proposed side-jets at launch for few moments
Only if you fired them most of the way up. If you only fire them for a few seconds, they're still dead weight for 90%+ of the booster flight, and the extra thrust wouldn't be enough to compensate.
> Isn't that going to look even cooler than what it is already?
Rocket design is not about looking cool, it's about shaving off as much weight as you can and adding as much thrust as you can. Cool looks are generally an afterthought (in fact, by going for an unpainted stainless steel, SpaceX is saving a lot of weight that would be added if they had to paint the entire thing).
SpaceX isn't building to cover ideal microoptimizations in certain parts of flight, they're building the best overall configuration that will work in the specific launch profile that they are targeting.