RFC Bug 425: changed expectation on textreplacewholetext07

I would appreciate review and feedback of bug 425 
(http://www.w3.org/Bugs/show_bug.cgi?id=425) originally reported by 
Andrew Clover regarding L3 Core test textreplacewholetext07.  My reading 
of the spec and the test agrees with Andrew that the original 
expectation of a NO_MODIFICATION_ERR is not correct.  I have committed 
changes to the CVS where the expectation of this test is rewritten and 
added an additional test (textreplacewholetext08) that should throw an 
NO_MODIFICATION_ERR.

textreplacewholetext07 fabricates a document similar to

<!DOCTYPE name [
<!ENTITY ent4 "<ent1element>foo</ent1element><?pi?>">
]>
<name>Margaret Martin&ent4;New Text</name>

Then attempts to use replaceWholeText on the text node containing 
"Margaret Martin" and expects a NO_MODIFICATION_ERR.  Since the ent4 
reference contains no logically-adjacent text, it should not be involved 
in the replaceWholeText operation and should not trigger the 
NO_MODIFICATION_ERR.

textreplacewholetext08 recreates the scenario from the spec and should 
result in a NO_MODIFICATION_ERR.

Xerces-J passed the original textreplacewholetext07 and fails both the 
new textreplacewholetext07 and 08.