Think of it like this, every kg added to Starship is one kg less to Orbit. And maybe like 10-30kg less to Mars.
For what you are talking about you likely need 16-32 SuperDracos, you need massive tank. A highly complex system of tubes along the vehicle and so on. You would reduce the payload to Orbit and return by a huge margin.
On earth, for a system to be reusable AT ALL, is incredibly hard. Like seriously, super hard. Their margins are already very thin. I'm not even sure what you suggest is possible at all.
And such a system would only prevent some very few cases. You already have engine redundancy. It wouldn't save you in many failure cases either. So its a gigantic increase in complexity for a slight reduction in potential failure.
The extra complexity quite apart from the performance would also lead to many new failure most that must be considered. If in a full risk analysis this would actually be safer is questionable.
If you want to increase safety adding a 4th Raptor would like by safer, as well as starting the flip earlier.
P.S: The fuel would also be highly toxic and would make operations with humans much more complex and much more expensive.
With standard staging this would be correct. With starship it's a bit different:
Every 1kg added to Starship is 1kg less [payload] to Orbit. And 1kg less to Mars. And maybe an extra tanker launch for refueling
But of course that inefficiency is gone be buffered by more refueling.
Still calculate the lost efficiency on the scale on a mars base and 1kg is a major.
Not to mention the system suggest above.
"SuperDraco engines use a storable propellant mixture of monomethylhydrazine fuel and dinitrogen tetroxide oxidizer"
It was one of the reasons why shuttles were so expensive to refly.
Hypergols are always highly reactive, by very nature of being hypergolic.
Also, for landing the engines need to be steerable which I think the emergency Superdraco ones aren't.
But I'm not sure that throwing extra engines into the scenario will help much. It takes very precise control to land a rocket. Those SuperDraco things are for going up fast, not coming down with sub-meter accuracy.
In fact many of the structural 'factors of safety' in airplanes are only 1.5, as opposed to a FOS of 8 in your car.
So again, it's not about just tacking on redundant systems, it's about getting the experience required to understand all the failure modes. The only times planes crash these days is when we encounter a new failure mode. Rocketry will get there as well.