You're right in being skeptical about all the "we must"s. It does seem, given how profoundly physical interventions can impact consciousness, that it is somehow physically implemented. And, therefore, it must be built out of something. But that does not mean that it's built out of elementary feelings.
As an analogy, take a limit cycle. This is a mathematical phenomenon where a time-varying process takes the same path repeatedly where various parameters change over time, but cyclically. If you perturb the system away from that limit cycle, it will be restored back. If consciousness was anything like this, it wouldn't be built out of "elementary feelings" any more than a limit cycle is built out of stationary attractors.
So to be honest, we must at least admit that we don't know for certain. We then may wish to entertain some educated guesses, and we may wish to consider what the minimal substrate would be for something that is sufficiently akin to what we experience as consciousness to deserve the same label. It does seem that some non-trivial machinery must be there to have not just a response to a perception but some secondary qualia associated with it. That it is like something seems to require both some memory and some pattern-matching capability or else the very notion of like cannot even make sense.
So it seems very likely that there is some minimal complexity below which consciousness, as we understand the term, does not exist.
However, you are wrong in suggesting that an amoeba could plausibly be as conscious as we are. An amoeba probably does not meet the minimal complexity required for consciousness, as it's doubtful that it has the requisite pattern-matching and memory capability. Furthermore, though it is not necessary that consciousness cause or regulate anything, one does at least suspect that if it were there, evolution would favor those organisms that tapped into that computation to modify behavior in ways that enhance survival. Thus we are justified in doubting (though not in concluding for certain) that animals like
C. elegans have consciousness. They simply do not behave in ways that are consistent with the kinds of capabilities that consciousness might add. (In particular, again, there is no obvious pattern matching.) Drosophila, however, do seem to have all the minimally requisite bits, so I think we should be more open-minded about whether they might have something like consciousness. (Personally I rather doubt it, but this is just a hunch based on how abstracted consciousness seems to me to be, and a further hunch about the effective depth of the fly's neural networks.)
Anyway, it's a difficult topic, and one where, sadly, people too often try to sweep all sorts of important and difficult concerns under the rug in order to try to make progress. I don't think this is progress--not on consciousness anyway--but rather misdirection (even if well-intentioned). I think that's what's going on in this case.