I would be (one of the) guilty there ... well it is a team effort. The original optimizing for WP43 using timers was done and tested by Dani R. some time ago and I further tested to get it optimized. Then ben.titmus re-engineered same WP43 code and I tested and optimized again.
Essentially you just try keep the processor asleep all the time (or at least as much as possible) and you limit the interrupt(ed) processor time to the minimum and during waking time, you make the processor do as little as possible. In terms of firmware base, C43 and WP43 have about 90% the same base loop system. C43 cannot be the same as it has a lot more for the processor to do when buttons are pressed - that is simply the "cost" of the different UI, drawing the lines on the menu, refreshing more and running the timers to monitor all the presses like fast, slow, double, triple and what not.
Monitoring of this stuff for optimizing, I use an oscilloscope to monitor the actual current drawn and we force activity pulses in various places in the code, which I then monitor using the oscilloscope, so the direct effect on the current consumption can be seen. If you really are interested in one of my testing methods, look at an older investigation report that we used internally - all issues identified in here have been fixed already (by Ben in Nov 2022).
Another very interesting issue is what to do if the battery is almost at it's end. The measurement issue can be compared with Heisenberg - even though I know that was meant way differently: The issue here is that you operate the processor to measure the battery voltage, while the processor is drawing current and therefore the voltage is dropping while you are measuring; the more your processor is not sleeping, and the flatter the battery is, the more unsure you are of what the voltage is; or was; or what it soon would be . If you are quick and only read the voltage and go to sleep again, you obtain the average of a low load or no-load voltage, which could create a false certainty of available power and when the user then asks the calculator to do some intense calculation like a program, the flat battery could arguably at this point not handle the surge in power and could die before you can shut down orderly. That would be a problem . So it is a complicated puzzle to solve. What's more, I am sure HP solved this especially for those venerable Voyagers which are as power clever as their Nasa namesakes with the radioisotope generators. I do think that the Nasa Voyagers may be the only things living longer on a battery than a 11C or 16C, but I digress.
I have not attended to the power issues since the above document, but at some point will return to it. In the meantime Michael (rightly so under protection of an NDA) provided me some essential DM42 hw information to help me with C43 improvements. My document shared above, clearly is pre-NDA though, so therefore shared for interest.