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

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}`