Improve `parse_paragraphs()`

[nils-herrmann]

Parsing the jats could be improved. Here are some observations:

1. Some documents (e.g. doi=10.1007/s40708-014-0001-z) use only <p> without attribute id.
2. Some documents put paragraphs inside lists:

<list list-type="order">
   <list-item>
       <p id="Par6">
         The normal traffic data is more than abnormal traffic for the training sample, and the data imbalance problem will lead to low precision.
      </p>
   </list-item>
<list-item>