No, it’s supposed to expose exactly a failure to appreciate the whole picture of the variations in genome size between species. Junk-DNA is a sort of null-hypothesis that explains that, and asking about onions, in comparison, can show how more specific hypotheses are really just ad-hoc and don’t make sense of the bigger picture.
The point is that if you think you’ve found some unique functional explanation why some species X has so much non-coding DNA, how does that explanation fare when you start comparing different species? If you think humans specifically need that much non-coding DNA to function, why would onions need five times more? And why would different species of onions, ostensibly similar in complexity and environment, need considerably different amounts?
Even very similar species can have very large differences in genome size, so it becomes very difficult for any one explanation for large amounts of non-coding DNA to account for the large variations there are. You’d have to come up with some unique explanation for each individual species, or at least a very large number of explanations for certain groups.
This generally makes for a very poor scientific explanation it becomes difficult to test, as you could in principle always assume that there’s some unique special circumstance with species X that explains why it needs that specific amount of non-coding DNA, and why another species needs a different amount - and that this special circumstance just doesn’t show up in your lab experiment.
The hypothesis that most (but not all) non-coding DNA is nonfunctional junk explains the large variations by considering what the nature of this junk-DNA is. It is mostly degraded transposons, pseudogenes, remnants of ancient viral infections, and various repetitive elements. Mechanistically this explains why it can expand in size in a relatively short amount of time.
And then there’s considerations from population genetics about mutational load in relationship to mutation rate, genome size, and population size, that explains why some species can carry lots of junk DNA, while others quickly get rid of it. It’s known as the drift-barrier hypothesis.
Think of prokaryotes vs large multicellular plants and animals. Prokaryotes have tiny streamlined genomes almost entirely void of transposons, and with comparatively little non-coding DNA (something in the 20-30% range of total genome size) and enormous population sizes, while on the other end animals and plants have giant genomes with almost the entirety being non-coding (and most of this being transposons) but much smaller effective population sizes.
I highly recommend watching this entire presentation by Michael Lynch on the drift barrier hypothesis in relation to the evolution of a host of cellular characteristics, ranging from genome size, organism and cell size, to metabolic and mutation rates:
This idea that the junk-DNA hypothesis is some sort of argument from ignorance (as if scientists can’t figure out what it does so they just give up and declare it nonfunctional junk because evolution demands it) is both factually and historically false. Extremely good evidence amassed over half a century of research has led to the conclusion that most non-coding DNA in multicellular eukaryotes is actually nonfunctional junk. People who think otherwise are simply misinformed.
Sort of… 
