Extend DEEP INSPECT model (SDK and Appliance) to allow user to send metadata to the application.

Today, one can create a Discover deep-inspect agent/application that executes operations on a Scale file system. As an example, there is an application available on github to collect the Scale user-defined attribute and assign it to a Discover tag. This fits the "collect metadata" paradigm. The request here is to enable a "use metadata" paradigm.

Today Discover's deep-inspect policy functionality (including the SDK's paradigm) uses this basic methodology (which I refer to above as "collect metadata" paradigm).
0. Agent/application is developed using the SDK.
1. User creates deep inspect policy associated with this agent. The policy definition includes list of metadata tags.
2. Discover node sends the agent: A) pathname list of file/object paths and B) tag list of metadata tags.
3. The agent collects values for each metadata tag in the "tag list," for each file/object in the "pathname list"
4. The agent sends Discover node list of files along with metadata tag:value pairs for that file.
5. Discover assigns the returned values to the metadata tags accordingly.

So, the limitation here is that the Discover node sends the agent ONLY the pathnames and the tag NAMES. No tag values are sent to the agent.

For the agent to assign VALUES to the user-defined attributes files of the Scale file system, the agent needs both the pathnames and the tag:value pairs. This is what I'm referring to as the "use metadata" paradigm - the agent in this example needs to "use" the metadata tags and their values in order to assign these values to the Scale user-defined attribute. (The assumption is that the Scale user-defined attribute name will be same as the Discover tag name, but that's an agent implementation detail.)

Note that if this - sending the metadata tags and their VALUES to the agent - were possible, then that agent can perform more complicated metadata extraction. The chargeback use case is good example. If the agent receives the pathname, the size, tier, and data source type (and whatever else is a factor), then the agent can perform calculations with those values that the Discover node cannot do. This includes multiplication, addition, if-then-else framework, etc. The results of the agent's calculations are then assigned to the "collect" tag(s) as it does today, sending them back to the Discover node for insertion into the database.