condense_multiline_comment throws exception when there are empty tags for Ecospold2 extractor

[tngTUDOR]

Current

The Ecospold2DataExtractor silently fails to parse multiline comment tags if they include empty tags.
Example source:

<?xml version="1.0" encoding="utf-8"?>
<ecoSpold xmlns="http://www.EcoInvent.org/EcoSpold02">
  <childActivityDataset>
    <activityDescription>
      <activity id="c40e3c0a-292f-45a5-88cd-ed18265cb7d7" activityNameId="06ffc039-f73b-440c-a5a0-8dc7393a5e71" parentActivityId="0fcbe58d-60b6-46d9-86bf-183d14732a4d" inheritanceDepth="-1" type="1" specialActivityType="0" energyValues="0">
        <activityName xml:lang="en">concrete block production</activityName>
        <includedActivitiesStart xml:lang="en">Includes start stuff</includedActivitiesStart>
        <includedActivitiesEnd xml:lang="en">Includes end stuff</includedActivitiesEnd>
        <generalComment>
          <text xml:lang="en" index="0">Things and stuff and whatnot</text>
          <text xml:lang="en" index="1">a Kikki comment</text>
          <text xml:lang="en" index="2"></text>
        </generalComment>
      </activity>
</xml>

The previous xmls snippet should produce the following string (when using the Ecospold2DataExtractor.condense_multilne_comment() method on it):

"Things and stuff and whatnot\na Kikki comment\n\n"

but it yields an empty string, producing the following diff in the tests:

-     'comment': 'Things and stuff and whatnot\n'
-     'Included activities end:  Includes some stuff\n'
+     'comment': 'Included activities end:  Includes some stuff\n'

Expected

The condense_multiline_comment method of the Ecospold2DataExtractor class should should process correctly "comment" sections in ecospold2 xml files, either by omitting empty "text" tags, or by adding a line break for such.