job management

[seungwookim]

for now we can request train with same network id and version multiple times at a same time, but this duplicated request need to be blocked for the performance and integrity issue

[seungwookim]

class to fix : api.views.RunManagerTrainRequest
class to add : cluster.common.job_manager.JobManager
how : add job management class on post and put action (check if same nn_id, ver request is already on working)