Add test to reproduce issue in impl Stream for Entries causing filename truncation

[RazerM]

This is from edera-dev/tokio-tar#3

---

I tracked down this issue in astral-sh/uv#5450 (comment)

https://github.com/edera-dev/tokio-tar/blob/4ee357285b5053e6bfada7f117e530b4da94b74a/src/archive.rs#L317

            if is_recognized_header && entry.header().entry_type().is_pax_local_extensions() {
                if self.pax_extensions.is_some() {
                    return Poll::Ready(Some(Err(other(
                        "two pax extensions entries describing \
                         the same member",
                    ))));
                }
                let mut ef = EntryFields::from(entry);
                let val = ready_err!(Pin::new(&mut ef).poll_read_all(cx));
                self.pax_extensions = Some(val);
                continue;
            }

if Pin::new(&mut ef).poll_read_all(cx) is Poll::Pending then ready_err! returns it, so the Pax extension is lost. The same would apply to a pending poll that occurs while a > longlink or longname is being prepared. When poll_next is called again the next entry header is parsed.

This PR demonstrates the issue by creating an AsyncRead impl which pends every second time it is polled.

Commenting out this line makes the test pass, because the reader doesn't enter a pending state in the "wrong" place.

It is probably also the cause of dignifiedquire/async-tar#39

[endbr64]

Please take a look at dignifiedquire/async-tar#55

[charliermarsh]

I applied dignifiedquire/async-tar#55 locally, but unfortunately the test is still failing for me.

[charliermarsh]

Ok, going to start looking into this more deeply since I've now determined that it's the cause of these spurious failures: astral-sh/uv#2235

[charliermarsh]

@endbr64 -- If you see https://github.com/astral-sh/tokio-tar/pull/40/files#diff-bc2127a92faf1e62dd213ca21bbd92e42da37f12597ab58ee95a1abcbdf25710R326, I had to make one additional change to your patch, which is: we need to store the entry that we didn't finish populating. Otherwise, we advance past it on the next iteration.

[charliermarsh]

I merged this into the wrong branch (oops). Will amend.