For the same reason
nobody else does 3A+2Lo: because the tiny base value increase isn't worth the extra difficulty. The judging system we have does not accurately reflect the difficultly of everything. It takes WAY more control to do a Loop combo on the end of a hard jump, particularly one that flows well. It's so much easier to just jam your toepick into the ice and do the toeloop.
If Hanyu had to train 3A+2Lo, which was a "necessity" for top men back in 1988 (!!!), then he would be able to do it beautifully I'm sure. Would he be able to do 3A+3Lo reliably if he actually put the time in to train it? I don't know for sure, but I do think out of all the guys, he's someone who would be most likely to master it. Given that this combination is very possibly the most realistic way Hanyu could find a technical advantage over Chen, it would be wise of him to pursue it. This is literally the most valuable a +3Loop combo has ever been under the CoP judging system, because of how it interacts with the number of jumping passes and attempted jumps being performed.
There's obvious reasons why that's not already happening.
It's strange though, you always assert that something isn't possible, as a straw-man argument.
Just a few years ago you said it wasn't feasible for someone to put 5 Triples in the second half of their program. 
Adam Rippon of all people was suddenly doing high quality 3Flip+3Loop after never doing combos like that before. So let's stop the tiresome argument about how such things are not realistic, simply because they haven't been done yet. The fact of the matter is, there are beneficial ideas out there that certain skaters and coaches haven't considered yet.