There's no such feature in EAGLE, and probably neither in other EDA-tools. And odds are low, something like this will be implemented in the near future.
Why that, you may ask. Why is there a possibility to have several symbols in one device, but not more than one package?
It's because this will break the design process. Let's assume your first use case, which still makes some sense in a way. You have a bunch of 100 nF capacitors to place. You want to connect P$1 to VCC and P$2 to GND on every one of them and have only one symbol in the schematic. This will somehow fit, but the drawbacks come up soon. So you placed a C in your schematic and connected the two nets. Automagically 15 packages arise in your layout, their pins already connected with rubber wires. You place and route.
First drawback: Your device needs to specify in your library, how many packages resembling your part it connects. So, if you want 14 C, you need an extra device. This will bloat your library by an extra dimension! You may say: "Ok, then let's have the same mechanism, as in the schematic with optional symbols, that must be invoked to be drawn."
I say, you can't compare that. Breaking a part into several symbols in the schematic serves a purpose in the design process. You want to have a description as exact as possible of all functions in the schematic, by maintaining maximum readability. This is, why you split a 144 QFP µC into a bunch of 8-pin ports and some control blocks and a power supply block. This allows you to keep the net lines short by moving the ADC-block of the controller over to — let's say — the analogue part of your schematic and have it even on a separate sheet finally.
What's with your capacitor resembling 1-15 capacitors in reality? Which value to display? 100 nF or 1400 nF, if you invoked 14 elements? In your schematic you can no longer show, that you have a capacitor next to each IC. And you also cannot calculate the electric behaviour of your capacitance for sure, which will break an export to any simulation tool based on schematics like spice.
So even if you were accepting this drawback. How could you implement your second use case? Having several filters on a bunch of ADC inputs? This forthright said makes no sense at all. If you have only one symbol in your schematic, how can you connect it to more nets than the pins it exhibits? Just impossible.
For your first usecase you may go the way of making yourself a blocking capacitor symbol with power pins instead of passive pins. These pins will wire themselves, so you can reduce drawing effort in your schematic. But this will not increase the readability of your schematic, as nobody — except you — will know, which net they connect.
One last thought. If you have so many similar parts with the same function, that you get problems in keeping the overview, you will have enough other parts to spread everything across more than one schematic sheet to sort out the functional blocks.
Remember: You need a schematic to keep the function of your circuit obvious. This is the sole purpose of a schematic. Everything else you can accomplish by editing a netlist in notepad.exe (or vi, which I prefer).