Try fix occasional benchmark file decompression failures

For longer running benchmarks we download an xz compressed archive
containing debug data for a certain kernel version. In recent times the
decompression has started failing in CI, for no apparent reason. Local
reproduction has failed.
Looking at the code, it seems possible that we experienced short write
during download and then discarded the unwritten rest. Switch from using
std::io::Write::write() to std::io::Write::write_all() in an attempt to
fix this issues.

Signed-off-by: Daniel Müller <deso@posteo.net>

Author: d-e-s-o

dev/build.rs

@@ -454,7 +454,7 @@ fn zip(files: &[PathBuf], dst: &Path) {
             let contents = read_file(file).unwrap();
             let path = file.strip_prefix(dst_dir).unwrap();
             let () = zip.start_file(path.to_str().unwrap(), options).unwrap();
-            let _count = zip.write(&contents).unwrap();
+            let _count = zip.write_all(&contents).unwrap();
         }
     }
 
@@ -919,7 +919,7 @@ fn download_multi_part(base_url: &reqwest::Url, part_count: usize, dst: &Path) {
     for part in 1..=part_count {
         let url = reqwest::Url::parse(&format!("{}.part{part}", base_url.as_str())).unwrap();
         let response = client.get(url).send().unwrap();
-        let _count = dst.write(&response.bytes().unwrap()).unwrap();
+        let _count = dst.write_all(&response.bytes().unwrap()).unwrap();
     }
 }