Fix text character truncation in py3 + bump to next version

- Fixes issue in Py3 when converting text characters to byte strings, but in Py3 converts to unicode instead, because uses the Py2 specific str() function, instead of the version neutral b(). When the text contains non-ascii 2-byte unicode values this results in truncating the unicode length instead of the byte length, and thus results in incorrectly padded byte lengths and data values ending up in the wrong field/column. See #157, and also #148. 
- Also bump to next version.

Author: karimbahgat

shapefile.py

@@ -2,12 +2,12 @@
 shapefile.py
 Provides read and write support for ESRI Shapefiles.
 author: jlawhead<at>geospatialpython.com
-date: 2017/08/24
-version: 1.2.12
+date: 2018/09/8
+version: 1.2.13
 Compatible with Python versions 2.7-3.x
 """
 
-__version__ = "1.2.12"
+__version__ = "1.2.13"
 
 from struct import pack, unpack, calcsize, error, Struct
 import os
@@ -989,8 +989,8 @@ def __dbfRecords(self):
                     else:
                         value = b(' ') # unknown is set to space
                 else:
-                    # anything else is forced to string
-                    value = str(value)[:size].ljust(size)
+                    # anything else is forced to byte string
+                    value = b(value)[:size].ljust(size)
                 if len(value) != size:
                     raise ShapefileException(
                         "Shapefile Writer unable to pack incorrect sized value"