The hard thing with taste is that it's usually the sum of many quantitative traits, each one being governed by several genes which can be located on different chromosomes, as seen in tomato[1]. Transferring enough of them into the modern genome through backcross would probably require a crazy amount of generations... or a crazy amount of luck :)
You would need to map the relevant loci and then use a speed congenic selection strategy. Done in mice easily. Mapping the loci modulating traits like disease resistance or taste is the hard part.
[1] https://doi.org/10.1093/jxb/erf058