Support for fine grain authorization in Pinot

This PR adds support for fine grain authorization for all REST APIs in Pinot.

What is fine grain authorization?

Essentially, every API access can be thought of performing some "action" on a specific resource id "targetId" of type "TargetType". Hence before executing the REST API, the implementation validates if the caller has required access to perform the action on the specific resource, and proceeds if has access, else returns 403 error.

How can it benefit Pinot users?

This framework provides ability to give access to perform one action on a specific table, or build higher level user profiles to give access to a set of actions. For instance, one user (or group or users) can be configured to execute SQL queries only, and other user can be allowed to perform admin actions for a specifc table. One such higher level access control in the industry is role-based access control (RBAC). RBAC restricts system access based on a person's role within an organization. Through RBAC, one can enforce granualr as well as broad policies.

Example of roles:

1. Cluster admin - allowed to perform any action on the cluster
2. Table admin - allowed to create table, drop table, rebalance etc. on one or a set of tables
3. Table reader - allowed to run SQL queries on one or a set of tables

How is fine grain authorization implemented?

There are 3 attributes required for granular authorization check:

1. Action being performed
2. Id of the resource on which the action is performed
3. Type of the resource on which the action is performed

Action and type of the resource is known statically for an API. However, the id of the resource (table name) can be in the path or query parameters of the API, or in the payload. For example: "/tables/{tableName}/instancePartitions" API has the tableName in the path parameter, and "/sql" has the table name in JSON payload. There are two annotations used for these scenarios:

1. Authorize - annotate the APIs to get the targetId from path or query params. The annotation has 3 elements:
   a. targetType - type of the target resource (Cluster or Table)
   b. paramName - path or query parameter to use to get the id or the resource
   c. action - to validate on the specific resource
   Request filter grabs the details from Authorize annotation to perform the validation.
2. ManualAuthorization - no authorization check by the request filter, and the check is performed in the API implementation once the resource id is available.

The implementation builds upon existing AccessControl interface for Broker and Controller APIs by adding two methods:

1. hasAccess - to check authorization using target resource type, resource id, and action
2. defaultAccess - this method is called by request filter if an API is neither annotated with Authorize nor ManualAuthorization to ensure no API calls are going though without any check.