Add supporting Semantic Convention documentation around Gauge metrics and Histograms

[adrielp]

Overview

During the process of contributing the VCS metrics there was a lengthy discussion around what type of metric some of them should be (either Gauge, Histogram, or this concept of a Gauge Histogram).

When adding another metric to the VCS list, we continued forward with the pattern of using gauge as the raw instrument type. As mentioned in this comment we should add some supporting documention in Semantic Conventions to better document when cases can be handled in this manner.

cc. @lmolkova