[Doc] Add a chart about the relative error of the percentiles aggregation.

docs/reference/search/aggregations/metrics/percentile-aggregation.asciidoc

@@ -146,6 +146,16 @@ the percentiles.  It is effectively trading accuracy for memory savings.  The
 exact level of inaccuracy is difficult to generalize, since it depends on your 
 data distribution and volume of data being aggregated
 
+The following chart shows the relative error on a uniform distribution depending
+on the number of collected values and the requested percentile:
+
+image:images/percentiles_error.png[]
+
+It shows how precision is better for extreme percentiles. The reason why error diminishes
+for large number of values is that the law of large numbers makes the distribution of
+values more and more uniform and the t-digest tree can do a better job at summarizing
+it. It would not be the case on more skewed distributions.
+
 ==== Compression
 
 Approximate algorithms must balance memory utilization with estimation accuracy.