Use ds names instead of ids when creating ds symlinks in generic mode

[sk4zuzu]

Predicting ds ids seems very inconvenient:

    SYSTEM_DS:
      system1:
        id: 100
        symlink:
          groups: [node]
          src: /mnt/nfs1/100/

Instead one-deploy could auto-create symlink pointing to the "id/" drectory with the name of the datastore that user provides (like "system1" would point to "100" directory), then it can be easily reused to create other symlinks without the need of knowing the actual id.