Gotcha. That looks like a slightly bigger chip, and otherwise almost identical in capabilities to the Tiny84, and should also work fine as you say.
Having extra pins is always nice; especially if you can leave those for the ICSP header so those don't have to do double duty -- making sure that re-programming the chip is possible, and doesn't drive your other circuit in some weird way, is important, too :-)
I've found that the best way to drive PWM for servos is to set the 16-bit timer to tick quickly (I use 2 ticks per microsecond,) and then calculate a "breakpoint" (in time) and "output value" (for the output register) based on the state of each of the pins. Do this by sorting the PWM duration of each pin from shortest to highest, and set the output register to 0xff minus the pins that should be low at that time.
- disable interrupts
- set the output register 0xff
- read the timer as a start value
- while the current timer minus start value < next-change-time, do nothing
- set the output register to the next-change-value, and increment the pointer to the next breakpoint
- if not yet at register value 0, go back to line 4
- enable interrupts
This will generate rock solid PWM with 0.5 microsecond resolution (much better than, say, the Arduino "Servo" library) for any number of output registers (you could drive more than one register in the loop.) The cost is disabling interrupts for up to 2 milliseconds every 30 milliseconds; if you need to poll some serial input port or something that can generate more data than can be buffered in that 2 millisecond time period, you're in trouble. That's why I like the USB hardware of the Atmegas; they buffer an entire packet :-)