Fix running traverse(): return 1

Else traverse() is short-circuited... which breaks the testsuite...
I wonder how the testsuite was working previous to rpm-4.17.0 which
fixed something that exposed that issue.
This test runs smoothly on mga[2-8]

Note for future: traverse() really should be documented...

2 files changed, 5 insertions, 0 deletions

diff --git a/RPM4/NEWS b/RPM4/NEWS
index f118539..5236b7c 100644
--- a/RPM4/NEWS
+++ b/RPM4/NEWS
@@ -1,3 +1,6 @@
+- Testsuite:
+  o Fix failure with rpm-4.17.0+
+
 Version 0.44 - 14 December 2020
 
 - Fix RPM4.xs/libperl.so mismatch on i586 by building with the flags used to
diff --git a/RPM4/t/05transaction.t b/RPM4/t/05transaction.t
index 218f4d0..b6c07c1 100644
--- a/RPM4/t/05transaction.t
+++ b/RPM4/t/05transaction.t
@@ -104,6 +104,7 @@ ok($ts->traverse(sub {
             $found++;
             (undef, $roffset) = ($hf, $offset);
         };
+        1;
     }), "Running traverse on DB");
 
 ok($found == 1, "The previously installed rpm is found");
@@ -127,6 +128,7 @@ ok($ts->traverse(sub {
             $found++;
             (undef, $roffset) = ($hf, $offset);
         };
+        1;
     }), "Running traverse");
 
 #ok($found == 0, "The previously removed rpm is not found");