Correct the <cmd> transformation for inline elements

[jhradilek]

When the <li> element only contained inline markup without any text around it, when converting it to a step, the dita-convert utility incorrectly used only the first element as <cmd> and moved the rest to <info>. This pull request corrects this behavior.

Example DITA input

<ol>
  <li>
    <ph>First phrase.</ph>
    <ph>Second phrase.</ph>
  </li>
</ol>

Incorrect DITA output

<steps>
  <step>
    <cmd>First phrase.</cmd>
    <info>
      <ph>Second phrase.</ph>
    </info>
  </step>
</steps>

Corrected DITA output

<steps>
  <step>
    <cmd>
      <ph>First phrase.</ph>
      <ph>Second phrase.</ph>
    </cmd>
  </step>
</steps>

Implementation checklist

• The code changes come with the corresponding test cases
• The code changes pass all tests (run python3 -m unittest in the project directory)
• The code changes come with appropriate documentation