The function of "Shift Register" is passing data between one iteration and the other, if you implement a conditional for "Shift Register" would be to select whether the data is passed or not. And if you do not pass this data, what value will make the next iteration?. Perhaps it would be enabled by a value "default", as it works when using the tunnel with the action "Use default if unwired".
Your idea is good just that the "Shift Register" if you wire externally, operates as a simple tunnel and the tunnel has already improved when used with the condition, so I see no point to it at some time used as "Shift Register "and other times as a tunnel, and with respect to the attached image, this function can be done with the" conditional tunnel ".
I can say that back when conditional tunnels were introduced, someone suggested the same idea elsewhere. I asked them to post it here and even came up with a sample graphic, but for some reason was unable to post it myself. I'm not sure why, but for some reason I have my doubts about this idea, even though it makes sense:
As you can see, this code calculates a simple factorial, but the bottom two options ignore the 3, so the result is a third of what it would normally be. Since feedback nodes are supposed to be basically the same as shift registers and they already have an enable terminal, it would make sense that SRs would have it to and that when it is false, the SR would keep the value it already has in it, but for some reason I still have my doubts.
National Instruments will not be implementing this idea. See the idea discussion thread for more information.