If the pom.xml of the plugin being tested contains special characters (for instance, accented characters in a developer name),] those characters are lost when writting the transformed pom.xml, and it may make it invalid.

If the original xml file uses utf-8, and contains an element like

   <name>José</name>

the resulting file keeps declaring utf-8 but it's actually written as ANSI, and the element appears as 

   <name>Jos�</name>

[Edit 20210712] - Removed reference to differences when using Java 11 that were actually caused by errors on my part while testing.