The only issue I could see with trying to keep the "gimmick" a secret is that most of this discussion will be in spoiler tags. Hopefully that's okay.
Also, this next one comin' up is... well, it's really really big.
Yeah, you can override one timed shift with another, or even one timed shift with itself. I'm hoping people won't accidentally "double-tap" the shifts too often, since it makes the timer sound effect play twice, and, well... it doesn't sound very good at all. There are a few other issues with the timer sound, as well - it continues to play even after the player dies, and if the timer is of the longest length, there's a good chance that it could interfere with them hearing another timer, so they're forced to either wait or deal with two simultaneous timer sounds. Similarly, even though the player can't actually carry shifts offscreen, the timer will still keep going even after the player leaves a given screen. Another issue I've noticed seems like it has to do with how many sound effects KS can handle at a time - with the longer timers, they'll occasionally get cut off either on the last "tick" (which is when the player teleports, so it's inconsequential) or two ticks beforehand (so I can't have situations where the player uses this long timer and then has to be in an extremely specific vertical spot, because that requires precise timing). There are a number of limitations and quirks, but the idea's so fun to work with and changes the pace of a KS level so uniquely that it's worth it, I feel. It's got a lot of the feeling of Monocromatica Grotta, but the player has more control of when to initiate that "rush" rather than being under constant pressure, so it's more suited for extended levels and thoughtful puzzles.
My application of the shift carry, at its core, is more focused on finding somewhere to use the shift, activating it, and then moving to that point. Observe, plan, execute is the rhythm that drives most of this level. Your idea - surviving a dangerous room - was something that I considered, but ultimately decided wasn't really something that would "gel" with the rest of the level design (and I doubted I could do very well, anyway). That said, your idea is likely a better fit for Knytt Stories' enemy design philosophy than what I'm doing. Since almost all enemies have randomized movement, attack patterns, attack spreads, or some mix of the three, I have to be very picky about which enemies I place into the level. An inconsistent challenge with the shift carry could be extremely frustrating, since the player doesn't have time to wait and adjust their plan. However, when the enemies are treated less as complex obstacles, and more as active threats that the player has to be constantly avoiding, this can work - randomized enemy design keeps the player on their toes, instead of memorizing a specific strategy which can get boring when you're waiting for something to go off.