spark-6106:Support user group mapping and groups in view, modify and admin acls

[colinmjj]

Add following configuration:
spark.security.group.acls.enable: true|false to enable/disable group acl check
spark.security.groupMappingProvider: to config the implement for getting groups for user,
the default value: org.apache.spark.util.ShellBasedUnixGroupsMapping

The group acls can be configured with: "spark.ui.view.acls", "spark.modify.acls", "spark.admin.acls", for example:
spark.ui.view.acls : user1,user2|group1,group2
The user list and group list will be splitted by '|'.

[AmplabJenkins]

Can one of the admins verify this patch?

[rxin]

I'm going to close this pull request. If this is still relevant and you are interested in pushing it forward, please open a new pull request. Thanks!

[alpivonka]

I would like to bring up an opportunity for re-use.
with in the mapred-site.xml (for our implementation) we are using the following to allow access to mapreduce logs ...etc..
Why reinvent the wheel, most often the users/groups for both MR and Spark would be the some or all of the same user/groups.
My suggestion is to create a common list/set of properties between MR and Spark for acls. Instead of maintaining two separate lists
mapreduce.cluster.acls.enabled = true
mapreduce.job.acl-view-job=mapred,hue,* group-name
mapreduce.job.acl-modify-job=mapred,hue,* group-name