Trait implementation

[lgalabru]

This PR is addressing #1168.

By doing so, we're introducing a mechanism for dynamically calling contracts while keeping our ability to run static analysis on contracts.
When a contract need to dispatch a call dynamically, it has to describe the behavior expected from the callee by defining a trait.

(define-trait can-transfer-tokens (
    (transfer-from? (principal principal uint) (response uint)))

(define-public (swap (token-a <can-transfer-tokens>) 
                     (amount-a uint) 
                     (owner-a principal)
                     (token-b <can-transfer-tokens>) 
                     (amount-b uint) 
                     (owner-b principal)))
     (begin
         (unwrap! (contract-call? token-a transfer-from? owner-a owner-b amount-a))
         (unwrap! (contract-call? token-b transfer-from? owner-b owner-a amount-b))))

Traits can be imported, aliased (for avoiding name collisions) and re-used:

(use-trait trait-a .contract-1.can-transfer-tokens) 

(define-public (pay-subscription (token-a <trait-a>))
    ...

Developers also have the option to explicitly declare that their contract is implementing a specific trait:

(impl-trait .contract-1.can-transfer-tokens) 

(define-fungible-token token)

(define-public (transfer-from? (sender principal) (recipient principal) (amount uint)) (response uint))
    (ft-transfer? token amount sender recipient))

In this case, the static analysis will throw if the contract is violating the specifications defined by the trait.

[jcnelson]

Thanks again for all your work on this @lgalabru! I think we still need to get some clarity on whether or not the system needs to be strict on requiring a contract to impl-trait, but I could be convinced. My only hesitation is from how it affects analysis and runtime cost. Maybe @kantai can comment?

[kantai]

Great stuff, @lgalabru!

I left comments on the issues that I think need to be addressed. Otherwise, I think we should open an issue for moving the VM's trait check to be checked at the top-level contract-call execution (i.e. on transaction execution).

[kantai]

Looks great, @lgalabru, thanks for addressing my feedback! I left one more comment, and otherwise LGTM!

I also opened an issue #1288 for moving the trait validation logic.

[jcnelson]

LGTM! Thanks @lgalabru! Just a small typo fix and you're good to go on my end.