Restructures the traits page

With a better definition given by @moritz Also restructures. Refs #1957

doc/Language/traits.pod6

@@ -4,15 +4,15 @@
 
 =SUBTITLE Compile-time specification of behavior made easy.
 
-In general, L<traits|https://en.wikipedia.org/wiki/Trait_(computer_programming)>
-are used to change the behavior of a class, by mixing in attributes and methods
-or changing the implementation. That kind of concept is implemented in Perl 6
-via L<roles|/language/glossary#Roles>; in this context, Perl 6 calls I<traits>
-specifically those that change the internal representation of the class or type.
+In Perl 6,  I<traits> are compiler hooks attached to objects and classes that
+modify their default behavior, functionality or representation. As such compiler
+hooks, they are defined in compile time, although they can be used in runtime.
+
+=head1 The X<C<is> trait|is (trait)>
 
 Traits are used extensively in L<native calls|/language/nativecall> to
-L<specify the
-representation|/language/nativecall#Specifying_the_native_representation> of the
+L<specify the representation|/language/nativecall#Specifying_the_native_representation>
+of the
 data structures that are going to be handled by the native functions via the
 C<is repr> suffix; at the same time, C<is native> is used for the routines that
 are actually implemented via native functions. These are the representations
@@ -67,13 +67,16 @@ methods, as above. However, trying to instantiate them this way: C<my
 $do-instantiate = don't-instantiate.new;> will yield the error C<You
 cannot create an instance of this type (don't-instantiate)>.
 
-=head2 Traits and routines
+=head3 C<is> on routines
 
-Traits can be added to the definition of methods and routines to
+The C<is> trait can be used on the definition of methods and routines to
 establish L<precedence|/functions#Precedence> and
-L<associativity|/language/functions#Associativity>. They act as a L<sub
-defined using C<trait_mod>|/type/Sub#Traits> which take as argument the
-types and names of the traits that are going to be added.
+L<associativity|/language/functions#Associativity>. They act as a L<sub defined
+using C<trait_mod>|/type/Sub#Traits> which take as argument the types and names
+of the traits that are going to be added. In the case of subroutines, traits
+would be a way of adding functionality which cuts accross class and role
+hierarchies, or can even be used to add behaviors to independently defined
+routines.
 
 =end pod