fix bytes_to_str bug (#747)

When we were using celery , and using `pickle` format to transfer a bytes variable like a file content, for example a picture file, then we will get an exception here. 
```UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte```
So add `errors='replace'` to avoid the encoding error here.

1 files changed, 1 insertions, 1 deletions

diff --git a/kombu/utils/encoding.py b/kombu/utils/encoding.py
index 93788bde..551cf5f3 100644
--- a/kombu/utils/encoding.py
+++ b/kombu/utils/encoding.py
@@ -53,7 +53,7 @@ if PY3:  # pragma: no cover
     def bytes_to_str(s):
         """Convert bytes to str."""
         if isinstance(s, bytes):
-            return s.decode()
+            return s.decode(errors='replace')
         return s
 
     def from_utf8(s, *args, **kwargs):