Add a section on performing structured clones from other specifications

This closes #855 by adding explicit guidance for those using structured
clone to ensure they properly set up entry and incumbent globals. It
also adds some additional helpful scaffolding, such as adding an
implicit conversion between Web IDL types and JS types.

source

@@ -8475,6 +8475,45 @@ http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C%21DOCTYPE%20html%3E..
    <li><p>Return ? <var>input</var>.[[<span>Transfer</span>]](<var>targetRealm</var>).</p></li>
   </ol>
 
+  <h4>Performing structured clones from other specifications</h4>
+
+  <p>Other specifications may use the <span>StructuredClone</span>,
+  <span>StructuredCloneWithTransfer</span>, and <span>IsTransferable</span> abstract operations (but
+  should not use the <span>TransferHelper</span> abstract operation).</p>
+
+  <p>In general, call sites may pass in Web IDL values instead of JavaScript values; this is to be
+  understood to perform an implicit <span data-x="concept-idl-convert">conversion</span> to the
+  JavaScript value before invoking these algorithms.</p>
+
+  <p>Call sites that are not invoked as a result of author code synchronously calling into a user
+  agent method must take care to properly <span>prepare to run script</span> and <span>prepare to
+  run a callback</span> before invoking these abstract operations, if they are being performed on
+  arbitrary objects. This is necessary because the <span>StructuredClone</span> operation can invoke
+  author-defined accessors as part of its final deep-cloning steps, and these accessors could call
+  into operations that rely on the <span data-x="concept-entry-everything">entry</span> and <span
+  data-x="concept-incumbent-everything">incumbent</span> concepts being properly set up.</p>
+
+  <p class="example"><code data-x="dom-window-postMessage">postMessage</code> performs
+  <span>StructuredCloneWithTransfer</span> on its arguments, but is careful to do so immediately,
+  inside the synchronous portion of its algorithm. Thus it is able to use the structured cloning
+  algorithms without needing to <span>prepare to run script</span> and <span>prepare to run a
+  callback</span>.</p>
+
+  <p class="example"><span data-x="url parser">Parsing</span> URLs can happen at a variety of times
+  that are not the synchronous result of author code. However, the URL parser only applies
+  <span>StructuredClone</span> to <code>Blob</code> objects, so it is not in danger of running
+  author code with incorrect <span data-x="concept-entry-everything">entry</span> and <span
+  data-x="concept-incumbent-everything">incumbent</span> values, and thus it too does not need to
+  perform these preparation steps.</p>
+
+  <p class="example">In contrast, a hypothetical API that used <span>StructuredClone</span> to
+  serialize some author-supplied object periodically, directly from a <span
+  data-x="concept-task">task</span> on the <span>event loop</span>, would need to ensure it performs
+  the appropriate preparations before calling into the structured clone algorithms. As of this time,
+  we know of no such APIs on the platform; usually it is simpler to perform the clone ahead of time,
+  as a synchronous consequence of author code, like <code
+  data-x="dom-window-postMessage">postMessage</code>.</p>
+
   <h4>Monkey patch for <code>Blob</code> and <code>FileList</code> objects</h4>
 
   <p class="&#x0058;&#x0058;&#x0058;">This monkey patch will be removed in due course. See <a