OK, now it really works with Cabinet files that are spread over more than one block (i.e. >32K in size).

The missing part came from a hint from Mark Adler, who advised:

Someone else was emailing me about this recently. They found out that each subsequent block uses the previous 32K block as a dictionary. You can use inflateSetDictionary() to set the dictionary for the next block before decompressing.

Also, update the unit test. I wasn't reading the khexedit display correctly.