i had a file with a single 0xB0 in it.
this char resolved to whatever might be but not to what it should.
as UTF-8 recommends and as it is commonly implement such a char should resolve to the illegal char marker U+FFFD.
A deeper check showed that indeed there is much fuzz that does not resolve as it should when using the test file from this location:
when using e.g. programmers notepad then its quite visible to me that the right edge line end markers are all one above the other. this is not the case for WinMerge and several other deviations do appear including the lack of use of the illegal char marker.
PS: in all test cases my install of WinMerge 220.127.116.11 Unicode showed UTF-8 on its bottom line