It does not seem to me that the reasons to save the drowning child could be ‘personal’ or self-regarding, and even if they could, they would be such that from them follow other imperatives that are at least seemingly other-regarding and on which the term ‘moral’ would be, I think, appropriate.
As for the scaling objection, it is a good one and one that has appeared in the comment section of my link-post on the EA forum. I will say here what I did there: that it seems very counter-intuitive to me to suppose there are no ‘rights’ and ‘wrongs’ and only things that are ‘better’ and ‘worse’, and that even if this is true, it would be useful to sometimes suppose the former exist, and distinguish between actions which come in the former category and those which come in the latter.
Yes, these are the self-regarding reasons I imagined you had in mind. My point stands, however, that the behaviour is at least seemingly other-regarding, and it is still action to which the term ‘moral’ appropriately applies. The kinds of things you are surmising about here are for the realm of meta-ethics and moral psychology; not normative and applied ethics. It might well be that I am only motivated by self-interest to act seemingly morally in accordance with consistency (crudely, that ‘egoism’ is true), but this says nothing as to what this moral system or what consistency requires.