Scheduling confusion

Hello,

I found some confusing behaviour in periodic job scheduling algorithm. I do not know if this is bug or feature. I just would like to share my thoughts regarding this.

Let me show you an example:

I have job that needs to be run every 6 seconds. The job itself takes about 2-4 seconds to complete. So, I'm expecting my job to be run every 6 seconds (if possible).

Unfortunately current algorithm doesn't takes into account job run time. So, first run is indeed scheduled after 6 seconds, but next run will be scheduled after job elapsed time plus 6 seconds. For example: job took 4 seconds to complete, so scheduler will scheduler next job run after 4 + 6 = 10 seconds.

Fortunately, if someone is interested I found quiet easy way to remedy my problem. We just need to slightly modify two methods of Job class:

First, method Job.run(self) should look like this (last_run is now time when job was started):

```
class Job(schedule.Job):
def run(self):
logger.info('Running job %s', self)
self.last_run = datetime.now()
ret = self.job_func()
self._schedule_next_run()
return ret
```

Second, we need to modify Job._schedule_next_run(self) by replacing line:
` self.next_run = datetime.datetime.now() + self.period`
with:
` self.next_run = self.last_run + self.period if self.last_run else datetime.now() + self.period`

Now, next job run will be scheduled like this: last_run + period and this is exactly what I was expecting. In case job elapsed time is longer that period, job will run as soon as possible.

I didn't tested this yet with jobs that needs to be run at given time but is should work.

Thanks!
Konrad