Prevent segfault for `COUNT(*)` with the `GROUP BY` hash map optimization

[RobinTF]

This is a bug I encountered when messing around queries like this

PREFIX p: <http://www.wikidata.org/prop/>
SELECT (COUNT(*) as ?c) WHERE {
  ?x p:P2860  ?z
}
GROUP BY ?x

would result in a segfault when hash map optimizations for group by are enabled because the code assumed exactly one child of any expression which isn't the case for COUNT(*)

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 89.45%. Comparing base (be816a9) to head (9019388). Report is 1 commits behind head on master.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #1471 +/- ## ========================================== + Coverage 89.41% 89.45% +0.03% ========================================== Files 346 346 Lines 25350 25353 +3 Branches 3381 3380 -1 ========================================== + Hits 22667 22679 +12 + Misses 1494 1491 -3 + Partials 1189 1183 -6 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[sonarcloud[bot]]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud