Ho ho ho!   I spent some time to try to resolve the remaining issues with PR 4682.  I think everything now works.  Returns from a final body of a try/finally was the remaining issue.  I've solved it by introducing the POP_NO_EXCEPT opcode.  I added a number of comments to help explain to future maintainers what is going on.

The PR still could use some polishing.  I have some test cases to add, to trigger the previously buggy try/finally + return case.  Also, I think the fblock_unwind_<block type> functions in the compiler could be done more simply.  I don't think we need a separate function for each type, just use a case statement on the fblock type.  That's a fairly minor detail though.

I would also like to add a bunch more comments.  I've spent many hours figuring out how all this stuff works, not that I totally understand everthing.   The test case test_contextlib test_exit_exception_with_existing_context() was especially brain-busting.  I'm a bit horrified that Python has become so complex to support such things.

Anyhow, I'm off for holidays so no time for further polish until after Christmas.