This is a very good question - and one that is often not fully understood outside the processing fraternity.  The basic answer is that it is because the grinding circuit in general has the highest capital cost and the highest operating costs (in terms of $/ton).  So it defines your overall production capacity.  Because of this, everything else is designed around the capacity of the grinding circuit in such a way that it doesn't take away from the grinding circuit capacity.  You don't want your crushing plant to be the limiting stage as this can interfere with the mining operation, and as a plant guy you don't want the mine guys screaming at you that they can't get their production numbers because your crusher is limiting their operation.  This is also why the sizing of the buffer between crushing and grinding is also critical.  The crushing plant has to absorb the fluctuations in mine production, while the grinding section is best operated at some steady state throughput.  If anything downstream of grinding (e.g. flotation, leaching, thickening, etc.) is limiting your throughput, then it is relatively cheap to add some more capacity, both in the design stage and as a retrofit in an operating plant.  Similarly, when you are operating the grinding circuit you want to have a control system to maximize throughput (if you are in a tonnage maximization operation - e.g. open pit copper flotation plant) or to match throughput with mine production (for example a low tonnage high grade gold mine) while maximizing grind (or whatever metric you use for grinding performance).  The bottom line is that because grinding is so expensive (both capex and opex) you want to make maximum use of the installed equipment (usually in terms of connected motor power) to maximize the economics of the overall operation.  I hope this helps.  Best Regards - Andrew