Better explanation about r.watershed vs. r.watersheds

src/raster/r.watersheds/r.watersheds.html

@@ -16,14 +16,15 @@ <h2>NOTES</h2>
 
 <em>r.watershed</em> (singular) can be used to create an input flow direction
 raster map. It can also create watersheds, but it uses an elevation map instead
-of a flow direction map, and does not take outlets as input.
-<em>r.watersheds</em> is more similar to <em>r.water.outlet</em> and
-<em>r.stream.basins</em>. Both modules take an input flow direction map from
-<em>r.watershed</em>, but <em>r.water.outlet</em> can delineate a watershed for
-one outlet point at a time and <em>r.stream.basins</em> is a sequential module
-for multiple watersheds. Unlike <em>r.stream.basins</em>, <em>r.watersheds</em>
-can use a column for watershed IDs, but using a non-default column is slower
-than using the default category (cat) column because of database queries.
+of a flow direction map, which is actually one of its outputs, and does not
+take outlets as input. <em>r.watersheds</em> is more similar to
+<em>r.water.outlet</em> and <em>r.stream.basins</em>. Both modules take an
+input flow direction map from <em>r.watershed</em>, but <em>r.water.outlet</em>
+can delineate a watershed for one outlet point at a time and
+<em>r.stream.basins</em> is a sequential module for multiple watersheds. Unlike
+<em>r.stream.basins</em>, <em>r.watersheds</em> can use a column for watershed
+IDs, but using a non-default column is slower than using the default category
+(cat) column because of database queries.
 
 <p>For comparisons, using an i7-1370P CPU with 64GB memory and a 30-meter flow
 direction map for the entire Texas (1.8 billion cells), <em>r.watersheds</em>