vendor: update to latest versions of everything

vendor/golang.org/x/net/html/const.go

@@ -97,8 +97,16 @@ func isSpecialElement(element *Node) bool {
 	switch element.Namespace {
 	case "", "html":
 		return isSpecialElementMap[element.Data]
+	case "math":
+		switch element.Data {
+		case "mi", "mo", "mn", "ms", "mtext", "annotation-xml":
+			return true
+		}
 	case "svg":
-		return element.Data == "foreignObject"
+		switch element.Data {
+		case "foreignObject", "desc", "title":
+			return true
+		}
 	}
 	return false
 }

vendor/golang.org/x/net/html/parse.go

@@ -470,6 +470,10 @@ func (p *parser) resetInsertionMode() {
 		case a.Table:
 			p.im = inTableIM
 		case a.Template:
+			// TODO: remove this divergence from the HTML5 spec.
+			if n.Namespace != "" {
+				continue
+			}
 			p.im = p.templateStack.top()
 		case a.Head:
 			// TODO: remove this divergence from the HTML5 spec.
@@ -1260,12 +1264,6 @@ func (p *parser) inBodyEndTagFormatting(tagAtom a.Atom) {
 		switch commonAncestor.DataAtom {
 		case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr:
 			p.fosterParent(lastNode)
-		case a.Template:
-			// TODO: remove namespace checking
-			if commonAncestor.Namespace == "html" {
-				commonAncestor = commonAncestor.LastChild
-			}
-			fallthrough
 		default:
 			commonAncestor.AppendChild(lastNode)
 		}

vendor/golang.org/x/net/http2/write.go

@@ -199,7 +199,7 @@ func (w *writeResHeaders) staysWithinBuffer(max int) bool {
 	// TODO: this is a common one. It'd be nice to return true
 	// here and get into the fast path if we could be clever and
 	// calculate the size fast enough, or at least a conservative
-	// uppper bound that usually fires. (Maybe if w.h and
+	// upper bound that usually fires. (Maybe if w.h and
 	// w.trailers are nil, so we don't need to enumerate it.)
 	// Otherwise I'm afraid that just calculating the length to
 	// answer this question would be slower than the ~2µs benefit.