Killing a job does not prevent successor jobs from being executed

Job Scheduler executes successor jobs in spite of the current job being killed. This is true for standalone jobs and for job chains, see the following samples:

a) The following job chain reproduces this problem:

<job_chain orders_recoverable="no"
visible="yes">
<job_chain_node state="start"
job="sample_1"
next_state="next"
error_state="error"/>
<job_chain_node state="next"
job="sample_2"
next_state="success"
error_state="error"/>
<job_chain_node.end state="success"/>
<job_chain_node.end state="error"/>
</job_chain>

<job order="yes">
<script language="shell">
<![CDATA[
echo job sample_1
c:\cygwin\bin\sleep 60
]]>
</script>
</job>

<job order="yes">
<script language="shell">
<![CDATA[
echo job sample_2
]]>
</script>
</job>

b) The following job reproduces this error:

<job>
<script language="shell">
<![CDATA[
echo job sample_1
c:\cygwin\bin\sleep 60
]]>
</script>
<commands on_exit_code="success">
<start_job job="sample_2"/>
</commands>
</job>

<job>
<script language="shell">
<![CDATA[
echo job sample_2
]]>
</script>
</job>

Whenever a job is killed by the built-in web interface or by an external kill command then this should be handled as if an exit code != 0 had occurred. This applies to jobs and job chains.