The period of your square wave is 100 us, but its duty cycle is 50%, so the input is at +5 V for only 50 us, not 100, before switching again.
This means it's only at +5 V for about 2.3 time constants rather than 4.6, and it's not able to come as close to fully charging the capacitor as it would have if kept at that voltage longer.
As another answer points out, you've also assumed a 5 V voltage step, while your simulation applies a 10 V step.
There's also a small effect because the simulator assumes the voltage source was at -5 V for a "very long time" before the simulation started, so there is a transient effect at the beginning of the simulation. You can see the 2nd peak in your output reaches a slightly higher voltage than the first one. You should let the simulation run for several (10?) cycles before measuring the peak voltage if you want a result close to the steady state behavior.