Document how the agent translates it's policy into a graph of started processes

With the V2 architecture changes and new people on the team ramping up on the agent it would be extremely helpful if documentation existed showing how an agent policy with multiple integrations installed were translated into a graph of running processes.

For example, consider the simplified agent policy below with two log inputs:

```yml
- id: logfile-system-IDA
name: system-1
type: logfile
use_output: default
data_stream:
namespace: default
streams:
- id: logfile-system.auth-IDA
data_stream:
dataset: system.auth
type: logs
paths:
- /var/log/auth.log*
processors:
- add_locale: null
- id: logfile-system-IDB
name: system-2
type: logfile
use_output: default
data_stream:
namespace: default
streams:
- id: logfile-system.auth-IDB
data_stream:
dataset: system.auth
type: logs
paths:
- /var/log/secure*
processors:
- decode_json_fields:
fields: ["inner"]
- id: logfile-system.syslog-IDB
data_stream:
dataset: system.syslog
type: logs
paths:
- /var/log/syslog*
```

It is not clear reading this that the two `logfile` inputs will be configured to run from a single filebeat process. The introduction of the https://github.com/elastic/elastic-agent-shipper further complicates this in that the filebeat process will be wired to a shipper instance per configured output.