As discussed in #545 we want to support running the collator relay chain node as its own process.
After #963 lands, we are in a position to run a parachain full node without running a relay chain full node internally. To run a proper parachain collator, we miss some of the subsystems usually running inside the full node. They do not offer an RPC interface, so I plan to start the relevant subsystems inside the collator.

Relevant Subsystems

After checking the code, I found that the following subsystems are involved:

• Availability Recovery
• Collation Generation
• Collator Protocol
• Network Bridge
• Runtime API
• Availability Store
• Chain API

The Availability store should be optional for now since the data we request from it will not be available inside the collator anyway and will then be requested from the Network Bridge.

Implementation Plan

• Implement a custom OverseerGen that will replace unused Subsystem with dummies that drain channels
• Feed a client to the Runtime API subsystem that will use the RPC Client introduced in Introduce rpc client for relay chain full node #963 to perform runtime calls

Related discussion in paritytech/polkadot#4763