I don't know of a clean way to remove this before it gets converted to XML. The best course of action is probably to increase the max record length and then remove <XMLElementToRemove> with the Field Remover processor.
If that is not possible, and the XML files are laid out one element per line, as in your question, it may be possible to use a separate pipeline to read in the files line-by-line as text and filter out any lines containing <XMLElementToRemove>.