Incorrent simple table parsing for RST->Markdown

[levitvas]

Failes to correctly transform and parse tables with empty columns when converting from RST to Markdown.
Here is a test table made in RST:

.. table:: Test Table

   ================================== ==== =========== =============== ===== === =====
   Functions for tests                1    2Test       3Untested       4Alg  5   6Col
   ================================== ==== =========== =============== ===== === =====
   a                                  yes              yes
   b                                       yes
   c                                                                         yes yes
   d                                  yes
   e                                                   yes             yes
   ================================== ==== =========== =============== ===== === =====

Base command used for running:
pandoc in.rst -f rst -t markdown+simple_tables+grid_tables -o out.md
Running this results in a grid table that is incorrect:
If ran with +simple_tables+grid_tables

+---------------------+-----+-------+-----------+------+-----+------+
| Functions for tests | 1   | 2Test | 3Untested | 4Alg | 5   | 6Col |
+=====================+=====+=======+===========+======+=====+======+
| a b c d e           | yes | yes   | yes       | yes  | yes | yes  |
|                     |     |       |           |      |     |      |
|                     | yes |       | yes       |      |     |      |
+---------------------+-----+-------+-----------+------+-----+------+

If ran with +simple_tables-grid_tables results in a regular table, but also incorrect:

<table>
<caption>Test Table</caption>
<thead>
<tr>
<th>Functions for tests</th>
<th>1</th>
<th>2Test</th>
<th>3Untested</th>
<th>4Alg</th>
<th>5</th>
<th>6Col</th>
</tr>
</thead>
<tbody>
<tr>
<td><p>a b c d e</p></td>
<td><p>yes</p>
<p>yes</p></td>
<td><p>yes</p></td>
<td><p>yes</p>
<p>yes</p></td>
<td><p>yes</p></td>
<td><p>yes</p></td>
<td><p>yes</p></td>
</tr>
</tbody>
</table>

How the table looks like in Markdown:

Test Table

Functions for tests	1	2Test	3Untested	4Alg	5	6Col
a b c d e	yes yes	yes	yes yes	yes	yes	yes

How it looks like in RST:

Pandoc version
Tested on latest version 3.3 on Windows.

[jgm]

Smaller test case:

========== ====
Function   1   
========== ====
a          yes 
b              
========== ====

It does depend on the empty cell; if that is filled then we get two rows.