DJM invokes shutdown listeners only sequentially

[KazankovMarch]

Example:

• SomeJob has many work-items
• SomeJob has WorkPoolRunningStrategy "Thread per work-item"
• during its shutdown listener invocation SomeJob waits for closing someResource with long timeout
• someResource has been broken - it can't be closed gracefully
• After rebalance worker tries to shutdown execution of SomeJob

It produce very long freeze because DJM shutdowns SomeJob executions sequentially and waits long timeout of closing someResource one time per work-item

[KazankovMarch]

ru.fix.distributed.job.manager.ScheduledJobExecution#shutdown
jobRuns.forEach(JobContext::shutdown);

[KazankovMarch]

I think we need to clarify that listener that added by JobContext#addShutdownListener should NOT to shutdown job in listener's Thread, but just send signal to job for closing. DJM waits job's termination by itself in ScheduledJobExecution#awaitTermination (during rebalance) and in Worker#awaitAndTerminate (during closing DJM)

[KazankovMarch]

after discussion we decided to clarify our API and give to user possibility to define threadPoolSize of executor in which we will launch shutdown listeners