[FEATURE] Create an XContent parser for Processors in pipelines.

[dbwiddis]

Is your feature request related to a problem?

Template parsing implemented in #47 assumes only one layer of nesting (e.g., a map object in Map<String, Object> will always be a Map<String, String>. This assumption isn't always true as shown in #44 where some manual content parsing was done.

What solution would you like?

Create an XContent parser for Processors in pipelines.

What alternatives have you considered?

Representing the processor JSON (below) in a string in the User Template. This would work but gets very ugly very fast as all the quotes need to be escaped.

Do you have any additional context?

All pipeline configuration formats follow a similar pattern :

{
    "description" : "",
    "processors": [
        {
             "processor1" : {
                 // required inputs
             }
         },
         {
             "processor2" : {
                 // required inputs
             }
         },
         ....
         ...
    ]
}

We can definitely generalize the creation of the configuration as long as the workflow data has a list of processors and their required inputs. The processors themselves vary in terms of format, so doing it this way would require some xcontent builder for each type of processor, since we'll need to write to specific fields.

Originally posted by @joshpalis in #44 (comment)