Fix for mangled text in Aux.csv

[gooosedev]

This PR should fix the generation of the mangled Aux.csv file (issue #81),

the issues seems to orginate from the conversion script
Inside the convert function when processing a element e with a ctype of 無活用型 leading to the last byte of the POS being truncated:

the solution that I've found was to add the following special case

else{
        # special case for some non inflecting.
        if ($ctype eq "無活用型"){
            my $result ="";
            for (@$surface) {
                $result .= sprintf ("%s,0,0,0,%s,%s,%s,*,%s,%s,%s\n",
                $_, $pos1, $pos2, $ctype, $_, $reading, $meaning);
            }
            return $result;
        }
    }

resulting in theses entries:

です,0,0,0,助動詞,*,無活用型,*,です,です,代表表記:です/です
まい,0,0,0,助動詞,*,無活用型,*,まい,まい,代表表記:まい/まい
ことだ,0,0,0,助動詞,*,無活用型,*,ことだ,ことだ,代表表記:ことだ/ことだ

---

note: the model.def, left-id.def and right-id.def will probably need to be re-generated and the model re-trained.