[bugfix] Fix division by zero when calculating the poll rate

[rsarm]

On my laptop quite often I get the error

[==========] Finished on Tue Sep 21 22:13:38 2021
Run report saved in '/Users/sarafael/.reframe/reports/run-report.json'
./bin/reframe: run session stopped: zero division error: float division by zero
./bin/reframe: Traceback (most recent call last):
  File "/Users/sarafael/git_/reframe/reframe/frontend/cli.py", line 1062, in main
    runner.runall(testcases, restored_cases)
  File "/Users/sarafael/git_/reframe/reframe/frontend/executors/__init__.py", line 431, in runall
    self._runall(testcases)
  File "/Users/sarafael/git_/reframe/reframe/frontend/executors/__init__.py", line 504, in _runall
    self._policy.exit()
  File "/Users/sarafael/git_/reframe/reframe/frontend/executors/policies.py", line 545, in exit
    self._pollctl.running_tasks(num_running).snooze()
  File "/Users/sarafael/git_/reframe/reframe/frontend/executors/policies.py", line 68, in snooze
    f'(current poll rate: {self._num_polls/t_elapsed} polls/s)'
ZeroDivisionError: float division by zero

[ekouts]

Lgtm. I agree with printing None in the log, but I am not sure if there is something more appropriate, like 0 or inf.

[vkarak]

It is very strange that t_elapsed can be zero! Could you reproduce that consistently?

[codecov-commenter]

Codecov Report

Merging #2199 (6d381ad) into master (59217e7) will increase coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head 6d381ad differs from pull request most recent head 1fef416. Consider uploading reports for the commit 1fef416 to get more accurate results

@@           Coverage Diff           @@
##           master    #2199   +/-   ##
=======================================
  Coverage   85.53%   85.53%           
=======================================
  Files          55       55           
  Lines        9743     9744    +1     
=======================================
+ Hits         8334     8335    +1     
  Misses       1409     1409           

Impacted Files	Coverage Δ
reframe/frontend/executors/policies.py	99.69% <100.00%> (+<0.01%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 59217e7...1fef416. Read the comment docs.

[rsarm]

I think it has to do with the precision that the time package operates with on different platforms. It looks like my laptop's precision is lower than Piz Daint's.

Laptop:

>>> import time
>>> time.time_ns()
1632381243676665000
>>> time.time_ns()
1632381244415043000
>>> time.time_ns()
1632381245058365000

Piz Daint:

>>> import time
>>> time.time_ns()
1632381235653486557
>>> time.time_ns()
1632381247922103471
>>> time.time_ns()
1632381422630926203