work around bug in lxml's incremental parser

lxml is applying basic UTF8 decoding to each chunk, which fails when a chunk ends in the middle of an UTF8 sequence

```
Traceback (most recent call last):
  File "legi/tar2sqlite.py", line 510, in <module>
    main()
  File "legi/tar2sqlite.py", line 486, in main
    process_archive(db, args.directory + '/' + archive_name)
  File "legi/tar2sqlite.py", line 262, in process_archive
    xml.feed(block)
  File "src/lxml/parser.pxi", line 1217, in lxml.etree._FeedParser.feed (src/lxml/lxml.etree.c:114563)
  File "src/lxml/parser.pxi", line 1339, in lxml.etree._FeedParser.feed (src/lxml/lxml.etree.c:114436)
  File "src/lxml/parser.pxi", line 586, in lxml.etree._ParserContext._handleParseResult (src/lxml/lxml.etree.c:105777)
  File "src/lxml/parser.pxi", line 595, in lxml.etree._ParserContext._handleParseResultDoc (src/lxml/lxml.etree.c:105896)
  File "src/lxml/parser.pxi", line 706, in lxml.etree._handleParseResult (src/lxml/lxml.etree.c:107604)
  File "src/lxml/parser.pxi", line 635, in lxml.etree._raiseParseError (src/lxml/lxml.etree.c:106458)
  File "<string>", line 227
lxml.etree.XMLSyntaxError: Input is not proper UTF-8, indicate encoding !
Bytes: 0xC3 EOF, line 227, column 289
```

legi/tar2sqlite.py

@@ -258,8 +258,7 @@ def count_one(k):
                     skipped += 1
                     continue
 
-            for block in entry.get_blocks():
-                xml.feed(block)
+            xml.feed(b''.join(entry.get_blocks()))
             root = xml.close()
             tag = root.tag
             meta = root.find('META')