Thanks for writing the test.

Yes, I did read the comment.  As I understood it, RETURN_VALUE is needed
only so that various optimization can assume codestr[] cannot suddenly
end without one.  E.g. if you match for BINARY_ADD, you can safely check
the next command: if BINARY_ADD matched, there is a _guaranteed_ next
command, not an out-of-array failure.

Such proposed fake RETURN_VALUE _must_ be unreachable, so it must not be
problematic at all.  If it was reachable, real codestr[] would end in
reachable non-return command, which must not happen during compilation.
 I dunno, maybe interpreter guards against execution point falling of
the bytecode string, but in any case this must not happen in
non-corrupted files generated by the bytecode compiler.