[monorepo] replace Transfer.sol mechanism with Interpreters

[ldct]

This PR refactors the contracts and node and fixes: #1222, #1157, #1442

The design is as follows. App Definition resolve functions now return unstructured data which is to be consumed by an "interpreter" contract which runs in the multisig context and handles state deposit payouts in dispute scenarios. App Definitions are classified by "resolve type", e.g.,

• TwoPartyOutcome: the app is some kind of 2-party game, and the resolve function returns a uint256 in the set {0, 1, 2} which means "player 1 won, player 2 won, draw" respectively
• ETHTransfer: the resolve functions returns a list of (to, value) tuples

State deposits constrain what apps they can be used with: for instance, a deposit of ETH can be used for apps of resolve type ETHTransfer or TwoPartyOutcome, but a deposit of two ERC721 NFTs can only be used for TwoPartyOutcome.

hard-coded assumptions

• virtual apps always resolve to type 2-party outcome

temporary hacks to be removed in the future

• A resolveType function is introduced in all app definitions. This is used by the uninstall protocol to determine how to update the free balance during uninstall.
• A limitOrTotal field is added to the AppInstance class. For app instances interpreted by ETHInterpreter, this represents a limit on the total ETH; for those interpreted by TwoPartyEthAsLump, this represents the total funds committed to the app.
• The ttt bot assumes that it is player 2
• interpreterHash is computed but not checked in StateChannelTransaction

[alxiong]

@IIIIllllIIIIllllIIIIllllIIIIllllIIIIll two extra comments:

1. better comments would help. (e.g. I don't really know what am I supposed to feed into resolution and interpreterParams
2. in the future, it would be more friendly if we provide a predefined set of interpreter options (i.e. preset) for most use case such that developers can just choose one (from an enum, which contains a list of common interpreters) by specifying a number, or they can customize their own for more complicated app.

in general, I do think current change is great and will support coinshuffle app requirement. Great work!

[ldct]

might be blocked on duaraghav8/Ethlint#261