Think of combinations of instruments playing in parallel unisons, fifths, and octaves as though they were kinds of composite instruments. Then an arranger writing for violin, flute, and cello wouldn't have just three instruments at his disposal, but also many more like the univiolinflute, the octaviolinflute, the suboctaviolinflute, the univiolincello, the octaviolincello, the doubloctabiolincello, the octaflutecello, the doubloctaflutecello, etc.
The reason for the rule against parallel octaves or unisons is that having groups of two or more instruments play parallel octaves or unisons will often make it sound as though one has switched to using a different set of instruments. This can be a good thing if done at places where such a switch would make musical sense, but bad if such switches seem to occur willy-nilly without rhyme or reason.
The goal of music theory is not to identify things that are "good" or "bad", but rather to allow composers to identify how things are likely to be perceived. If a composer wants part of a phrase to sound as though it's produced by a univiolinflute while other parts are played by a distinct violin and flute, great--use parallel intervals to achieve that effect. The music theory rules against parallel intervals doesn't say such things are "bad", but rather say that parallel intervals are likely to create such an effect whether the composer wants it or not, and composers should avoid them except when that effect is wanted.