Project organization plan

Current Foundry's project structure is somewhat tangled.
I suggest following new structure.

1. We have a separate repo for types/traits/utilities widely used in the host and modules. (Let's say `foundry-suite`) Some types might have their `async-graphql` implementations. The module author will import both `module-rt` and `foundry-suite`.
Also, the coordinator - module interfaces are included here as well.

2. App descriptor is separated into two independent parts: Foundry configuration (genesis config, transaction services) and general linking instructions (export, import, init config). Let's say the latter as a 'link descriptor'.

3. We keep the module system (Mold) pure, not involved in any blockchain context. It will be only about sandbox, links, communication and bootstrapping. Here are the components included
- RTO
- `module-rt`
- sandbox implementations
- link-desc (see section 2)
- weaver
- `codechain-module` (linkable, linker, Port, ...)

4. Coordinator will be just one use case of Mold, with some predefined services to import and export (coordinator - module interfaces).

5. We publish Mold as a general framework, which can be used widely out of blockchain context.

Doing this, we can have three clearly separated groups.

1. The consensus engine: Anything below the engine interface. DB, Tendermint, P2P, GraphQL server and so on. Current `foundry` repo will be about them.

2. The general module system: As described above.

3. The bridge between 1. and 2. : Coordinator, types, predefined services, utilities, ...

Any comments are welcome.