Im not 100% certain on this, but i believe its because you need to get certain values to achieve the needed value, but don't have availability of the exact value, but can achieve it by using small bypass caps in parallel with the main caps.
Like say you need 9.1mH value capacitor, but none are available in the brand you want. So you run a 9mH & 0.1mH in parallel to get the same effect. (Im no expert, but that's the kind of sense it makes to me conceptually)
Capacitor values are measured in microfarads, uf. It's coils (inductors) that are measured in millihenries, mH.
The reason for the small bypass capacitor in parallel with a large one is to shunt high frequency noise around the main capacitor so that it doesn't affect the signal.
Adding a 0.1uf across a 9uf filter capacitor will effectively change the total to 9.1uf, but that is not important. It represents a tuning difference of only 1%, which is inconsequential, and well within the tolerance band of most filter capacitors. It is the ability of the small cap to pass high frequencies around the large one that is important. Ergo the name, bypass capacitor.