C0 to DF take one continuation byte; E0 to EF take two.
It's invalid UTF-8 anyway, but at least this is what the test row meant:
overlong sequence with 3 bytes of what should have been two.
This updates the comment to match the character that we were actually
testing.
Change-Id: I85a8bd6da2c44f52b4e3fffd14a75df2600487aa
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>