[ML] Make it easier to map the autodetect process to Job ID

[davidkyle]

When running multiple anomaly detector jobs it can be useful to know which autodetect process relates to which job. The same could also applies to Data Frame Analytics. Knowing the job ID would be useful when collecting runtime metrics for each job such as the amount of memory the process is using.

Pytorch inference has a modelid argument that is parsed and never used but serves as a useful marker, similarly autodetect could have a job_id argument.

It is possible to extract the job Id from the named pipe filename which is for the log pipe is constructed as

--logPipe=/path/to/autodetect_{JOB_ID}_log_{PPID}