improve: just use content_LT to fix 16458

Just replacing element with content_LT, it works.

See #19522 (comment)

Author: i10416

compiler/src/dotty/tools/dotc/parsing/xml/MarkupParserCommon.scala

@@ -202,13 +202,6 @@ private[dotty] trait MarkupParserCommon {
   /** skip optional space S? */
   def xSpaceOpt(): Unit = while (isSpace(ch) && !eof) nextch()
 
-  /** skip optional space S? and return the number of consumed characters */
-  def xSpaceOptN(): Int =
-    var i = 0
-    while (isSpace(ch) && !eof) do
-      nextch()
-      i += 1
-    i
 
   /** scan [3] S ::= (#x20 | #x9 | #xD | #xA)+ */
   def xSpace(): Unit =

compiler/src/dotty/tools/dotc/parsing/xml/MarkupParsers.scala

@@ -371,17 +371,10 @@ object MarkupParsers {
         // parse more XML?
         if (charComingAfter(xSpaceOpt()) == '<') {
           while {
-            if xSpaceOptN() == 0 then
-              nextch()
-              if content_LT(ts) then // Is `</>` valid xml?
-                xToken("/>")
-              charComingAfter(xSpaceOpt()) == '<'
-            else
-              // this is surely not a special node as any special node 
-              // should start with `<{special symbol}` without space.
-              nextch()
-              ts.append(element)
-              charComingAfter(xSpaceOpt()) == '<'
+            xSpaceOpt()
+            nextch()
+            content_LT(ts)
+            charComingAfter(xSpaceOpt()) == '<'
           } do ()
           handle.makeXMLseq(Span(start, curOffset, start), ts)
         }