Did anybody ever really think it was? Or is this just a strawman put up by people who are opposed to IQ rankings on either personal (waah, I didn't score that high) or political (waah, group X doesn't score very high on average) grounds?
The idea of general intelligence makes about as much sense as the idea of general physical fitness. That is, quite a bit. It's useless to ask whether [famous basketballer] is more or less fit than [famous footballer], but we can usefully answer the question of whether [famous footballer] is more or less fit than [random dude].
Similarly, if we're an organisation that wants to select people for physical fitness (like, say, the US marines) we could make up a test and a score (http://en.wikipedia.org/wiki/United_States_Marine_Corps_Phys...) which in some way quantify an individual's physical fitness with a numerical score. The tests used are somewhat arbitrary and the precise ordering you get on this scale is different to the ordering you'd get if you used a different though equally sensible set of tests, but this doesn't change the fact that "physical fitness" is a real thing and that this test is a reasonable way of quantifying it. All reasonable measures of general physical fitness will give highly correlated results.
2. As this article points out, almost half of the performance of the various subsystems does seem to have a common source. So getting rid of IQ/g completely wouldn't make sense.
Now, pointing out specifically which other factors are most relevant for specific tasks would be useful (especially for things that require reading emotions, since that's least correlated). Finding which factors are least correlated is useful (so you know what's most important to check tasks for).
Even without the research, quantifying what counts as "general aptitude" and what doesn't is hard to do, and measuring whatever we decide to quantify is even harder. Here's to hoping that a more nuanced view of intelligences take off :)