You seem to expect that level from Hollis. "Reasonably practicable" is a hugely vague statement. This unit is impossible to put together incorrectly and still have it actually pass all tests, correct? Testing is part of the assembly sequence, correct? From my perspective and definition of "reasonably practicable" Hollis went well above and beyond. Could more be done? Probably. Should it? Depends on who you ask. I mean, people WITH the unit didn't think it was possible to assemble it that incorrectly.
As for the CO2 sensor in the mouthpiece, it seems unreasonable to me to expect a CO2 sensor to ignore "normal" high CO2 levels upon exhale but expect "normal" low-but-nonzero levels upon inhale and actually assume it's correct. Plus, a CO2 sensor in the mouthpiece would be a miserable feat of engineering, and you could never change mouthpieces or, depending on the design, the loop hoses. It simply doesn't make sense to me to introduce another breach right there. Plus, the sensor could false-trip upon high exertion levels or incorrectly closing the DSV. It would make MUCH more sense to put the CO2 sensor right next to the rest of the sensors, after the scrubber, right where it's assumed to have an actual zero-reading so that any elevated reading sends alerts. All CO2-monitored 'breathers I've seen (admittedly not the best person to make this statement) have them in the head, a location that the deceased's unit cleverly circumvented.