Collapsing gas clouds fragment into multiple cores because the Jeans mass, that determines the minimum mass that becomes gravitationally unstable to collapse, becomes smaller if the cloud is able to contract without heating up too much. i.e.
$$M_J \propto T^{3/2} \rho^{-1/2},$$
where $\rho$ is the cloud density. Thus if the cloud density can increase but the temperature stays constant(ish), then the Jeans mass shrinks and the cloud becomes unstable to further fragmentation.
Multiple systems with $n>2$ are inherently unstable unless they are hierarchical. i.e. a star in a wide orbit around a close pair can be stable, as could two close binary systems orbiting each other. The condition for stability is roughly that the separation of the wider star must be 5-10 times that of the inner pair ([dependent on mass ratios and eccentricities - Eggleton & Kiseleva 1995). In most other cases, what happens is the ejection of other stars from a multiple system, leaving behind a binary system Durisen et al. (2001).
What makes a binary stable? We'll what would make it unstable? I'm not sure why you think they should spiral in towards each other. This can't happen unless there is some dissipative mechanism, like tidal interactions. Gravitational waves are ineffective in all but close binaries involving compact stellar remnants. They can't "break away" because they are gravitationally bound.
EDIT: Note that it isn't true that most systems are binaries. The majority of "systems" are in fact single stars. The binary frequency for solar-type stars is about 50% - i.e. as many singles as binaries; but the binary frequency for the much more numerous M-dwarfs is probably around 30% and so single stars outnumber binary systems, though it is not clear whether that is true at birth Duchene & Kraus (2013).