The problem is that sidecan gives us the direct line distance to the feature. From that, we can work out the horizontal distance, if we know the depth, or we could potentially work out the depth, if we knew the horizontal distance. In practice, slant-range correction (essentially, working out the horizontal distance) is done by using the depth under the boat and the straight-line distance to the feature, using simple trigonometry. Of course this can give rise to errors when the feature in question is at a different depth but it's all we have.
Multi-beam sonar, on the other hand, uses an array of narrow cone-angle beams, and this is what is generally used for professional surveying. This can give accurate, high density depths over a broad swath. The only consumer grade multi-beam sonar I have heard of was a Humminbird unit, which has now been discontinued for quite a while..