fix: batch collect jobs for scancel (snakemake#2114)

### Description

When using --slurm, exiting snakemake by SIGINT or ctrl-c helpfully
cancels spawned jobs. However, this is quite unreliable as it often
hangs for a few minutes and exits without canceling the jobs in the end
(without indicating that it failed either).

Slurm documentation on
[scancel](https://slurm.schedmd.com/scancel.html#SECTION_PERFORMANCE)
notes that a large number of scancel calls at the same time may result
in denial of service.

Snakemake runs scancel on each job
[individually](https://github.com/snakemake/snakemake/blob/main/snakemake/executors/slurm/slurm_submit.py#L136).
Instead, job ids should be collected and cancelled all at once.

fixes snakemake#2113

### QC
* [X] The PR contains a test case for the changes or the changes are
already covered by an existing test case.
* [X] The documentation (`docs/`) is updated to reflect the changes or
this is not necessary (e.g. if the change does neither modify the
language nor the behavior or functionalities of Snakemake).

Co-authored-by: Johannes Köster <johannes.koester@tu-dortmund.de>

snakemake/executors/slurm/slurm_submit.py

@@ -134,22 +134,23 @@ def additional_general_args(self):
         return [" --slurm-jobstep", "--jobs 1"]
 
     def cancel(self):
-        for job in self.active_jobs:
-            jobid = job.jobid
+        # Jobs are collected to reduce load on slurmctld
+        jobids = " ".join([job.jobid for job in self.active_jobs])
+        if len(jobids) > 0:
             try:
                 # timeout set to 60, because a scheduler cycle usually is
                 # about 30 sec, but can be longer in extreme cases.
                 # Under 'normal' circumstances, 'scancel' is executed in
                 # virtually no time.
                 subprocess.check_output(
-                    f"scancel {jobid}",
+                    f"scancel {jobids}",
                     text=True,
                     shell=True,
                     timeout=60,
                     stderr=subprocess.PIPE,
                 )
             except subprocess.TimeoutExpired:
-                logger.warning(f"Unable to cancel job {jobid} within a minute.")
+                logger.warning(f"Unable to cancel jobs within a minute.")
         self.shutdown()
 
     def get_account_arg(self, job):