[ML] Improve processing of job groups in data recognizer wizard

[peteharverson]

Currently when creating anomaly detection jobs from a module in the data recognizer wizard, each job ends up with all of the groups referenced across all the jobs in the module. The processing of the groups should be changed so that only the group(s) assigned in the job configuration file are added to the job, with an extra control added to the wizard to optionally override the groups defined in the module so that, for example, an extra group can be added to the jobs.

For example, in the siem_auditbeat module, the linux_anomalous_process_all_hosts_ecs job is put into groups

  "groups": [
    "siem",
    "auditbeat",
    "process"
  ],

in the job configuration file, but after creation of the job in the wizard, is also placed in the network group which is used by other jobs in the module:

[elasticmachine]

Pinging @elastic/ml-ui (:ml)