I think the term "source" is tripping you up. From a certain perspective, a microcontroller is a current source, in the sense that it is passing current between its two supply terminals - from the perspective of the negative supply, the IC is indeed a source of current from a higher potential. From the perspective of the positive supply terminal, the IC is sinking current from that positive supply, down towards a lower potential (ground).
Engineers can throw the term "current source" around rather recklessly, knowing that other engineers are aware that everything passing current is either a source or a sink of current, from one perspective or another.
The context here, though, is that the IC represents a path for current to flow from one supply node to the other. This context also implies that the microcontroller IC itself is responsible for determining the amount of current that it passes. Every time a push-pull transistor pair (a digital gate) inside the IC switches state there's a significant but short-lived spike of current. There are thousands of them inside a microcontroller, all potentially changing state thousands, or millions of times per second. Each transition is a sudden burst of current, thousands or millions of them each second.
Even the loads on an IO (output) pin produce current transitions, every time that output changes state - an LED connected to one of the microcontroller outputs might be PWM controlled, switching between full current (say 5mA) and no current at all, thousands of times each second.
Each current transition is fast, with many harmonics at many megahertz, and the loop around which those changing currents must travel is an antenna. The trick is to make that loop as small as possible, to minimise emissions, and to alleviate the responsibility of conduction of those currents from the ground and power planes at large - hence local energy sources (capacitors) to take on that responsibility, and small current loops through the capacitors, rather than around the entire main supply loop.
Every single element (such as an LED, or microcontroller) in those loops can be considered a "current source", or sink; the LEDs are either 5mA sources/sinks or 0mA sources/sinks, but it's easier to call them just "sources", and let the reader decide which, according to whatever model and perspective he has in mind at the time. It is a bit lazy, but we know what it means.
One usually associates a current or voltage source as a source of energy (hence the name), but this is not always the case - you can charge a battery (voltage source) by pushing current the "wrong way" through it, changing the sign of the power product \$P=I\times V\$. Similarly you can deliver energy to a current source by changing the polarity of voltage across it. Again, the name "source" doesn't really tell the whole story, but if an engineer had to explain the distinction, in terms of power, or current direction every time, he'd quit - it's much easier to say "source", in every instance, and let the reader infer the meaning from context.
In the context of this discussion, which is strictly current, the author seems to be implying that these elements are passing current from +5V (or +3.3V) to 0V (ground), and are to a large degree responsible for the magnitude of that current, which if you think about it, is exactly what a current source does. It's clearly not a current source in the strict academic sense, though, so I understand your confusion.
Personally I would not have called it a current source. I would have said that it is modulating current at very high frequency. It's still true though that the positive supply node sees that current-modulating element as an "AC current sink", and the ground node would see it as a "AC current source".
