Most of those failures are from abuse of backquote in defmacro-enhance which is in turn used by other systems.

If not due directly to defmacro-enhance, then due to similar ideas which are espoused by the author of (the text) let-over-lambda in which he seems to promote use of ideas that are easy to implement unportably but tricky to implement portably.
defmacro-enhance lacks a self-test that exhibits the problem, which is why that system by itself seems to pass.
The author of defmacro-enhance is aware of the issue. I never cc'd authors of the dependent systems, though I might do that if a fix is not forthcoming.

I'll take a look at nuclblog as I don't recall that showing up on my watch list.