Clarify that foreign content is not an insertion mode

The rules for parsing tokens in foreign content are selected per token by
the tree construction dispatcher and leave the insertion mode unchanged.

Follow-up to #12658. Fixes #4015.

Author: annevk

source

@@ -141092,6 +141092,11 @@ dictionary <dfn dictionary>StorageEventInit</dfn> : <span>EventInit</span> {
   <p>The <dfn>insertion mode</dfn> is a state variable that controls the primary operation of the
   tree construction stage.</p>
 
+  <p class="note">The rules for parsing tokens <span data-x="insertion mode: in foreign content">in
+  foreign content</span> are not an <span>insertion mode</span>. The <span>tree construction
+  dispatcher</span> selects them independently of the current <span>insertion mode</span>, which is
+  left unchanged while they are in use.</p>
+
   <p>Initially, the <span>insertion mode</span> is "<span data-x="insertion mode:
   initial">initial</span>". It can change to "<span data-x="insertion mode: before html">before
   html</span>", "<span data-x="insertion mode: before head">before head</span>", "<span
@@ -143793,8 +143798,8 @@ dictionary <dfn dictionary>StorageEventInit</dfn> : <span>EventInit</span> {
   </div>
 
   <p class="note">U+0000 NULL characters are handled in the tree construction stage, as part of the
-  <span data-x="insertion mode: in foreign content">in foreign content</span> insertion mode, which
-  is the only place where CDATA sections can appear.</p>
+  rules for parsing tokens <span data-x="insertion mode: in foreign content">in foreign
+  content</span>, which is the only place where CDATA sections can appear.</p>
 
   <h5><dfn>CDATA section bracket state</dfn></h5>