I've looked at many of the actual studies, although not the recent one you mentioned. I agree with your overall analysis, but would add one addendum - there are different types of happiness. The delineation I find most applicable here would be Daniel Kahneman's. He suggests that there are two types of happiness - experiential and remembering.

Experiential is measured by his day reconstruction method, as well as the experience sampling method mentioned by benthamite. Call it hedonic, moment by moment happiness.

Remembering is as it suggests - how we feel when we remember our past. This is why meaning has importance - we like to feel we've done things with our lives (e.g. he suggests this is why we try to fill our lives with 'memorable' events, even when these events themselves do not create the largest positive affect at the time).

As you've said, the papers are mixed on the experiential happiness side - some papers (self-report + experience sampling + day reconstruction) vote positive, others negative. On the remembering side, all papers I've seen have reported increases. What does this mean? I have no idea - the problem is that happiness is poorly understood.

Usually this is not a problem - most decisions lean quiet clearly in one direction or the other - that is, clearly increase happiness or clearly don't. What does it mean if parents report lower life satisfaction, but higher meaning in life? No idea.

But as the research stands now, I personally would not have children. Consider the effort required to raise children. Take that same effort and apply it to other areas of your life, where the happiness research is more clear, and your return on investment will be much higher (by an order of magnitude, given the effort required to raise children).