Address issue identifying of CDATA

The XML library wasn't identifying CDATA and skipping over it.
This was found when CDATA contained a <. This patch addresses
that issue.

core/encoding/xml/xml_reader.odin

@@ -308,6 +308,7 @@ parse_bytes :: proc(data: []u8, options := DEFAULT_OPTIONS, path := "", error_ha
 				case .Open_Bracket:
 					// This could be a CDATA tag part of a tag's body. Unread the `<![`
 					t.offset -= 3
+					t.read_offset = t.offset
 
 					// Instead of calling `parse_body` here, we could also `continue loop`
 					// and fall through to the `case:` at the bottom of the outer loop.
@@ -625,4 +626,4 @@ new_element :: proc(doc: ^Document) -> (id: Element_ID) {
 	cur := doc.element_count
 	doc.element_count += 1
 	return cur
-}
+}