1. The mathematical statement says,that, if the input is "bounded" that is within finite values ,then the output of the system, is also bounded. for a stable system.
2.In real systems , it is conceivable that the system is stable but the equipment fails if any component exceeds its specification. For example, in even a simple operational amplifier, if resistors are not chosen for proper heat dissipation , it might fail ,even though the circuit values assure a stable. response from a control point of view.. One other example is power system stability , where we encounter a breakdown if the torque angle exceeds a critical value during disturbances..
3 Another possibility is that the given definition ,could include limit cycles and attractors in chaos situations.
4 A possible application of summing n terms of input and output would be in Power system stability . Consider a sudden increase of load . in a power network. If we are able to sum power input and output at each node and impose some thresholds we can come up with a criteria for ensuring stable operation in real time.
For a linear system, stability is usually defined in terms of the "bounded input/ bounded output" (BIBO) criteria.If a system is BIBO stable, then the output will be bounded for every input to the system that is bounded.
Based on our control systems , we know that a transfer function is a ratio of OP/IP . Be it OP or IP they are characterized by a differential equation . The solution to such a differential equation may end up have be infinity, the response needs to be finite and measurable , hence BIBO .
Denominator largely effect stability (1/0 ) , in the freq-domain (or s) it is the pole that give stability points (graphically)
In the time domain we convolve input and system definition or functioning ( simple is impulse ) to get the overall plant performance . This again needs to be within the limit to characterize it as time-invariant.
Graphically , summing would give you the area under the curve .
Electrical systems : we have many many limitations such as bandwidth / overshooting /undershooting limits ...