Cross Shard Locking Scheme - (1)

[musalbas]

@vbuterin

This keeps the model of what cross-shard knowledge is required for executors to have the same as it was before, and pushes the work of actually performing the cross-shard operation to users.

This is similar to the method used in OmniLedger. This works well for applications where the user is locking a state that only they can modify, such as spending a UTXO, because if they lock a UTXO that belongs to them, but then do nothing, then the user is only harming themselves. But if the user locks an object that can be accessed by anyone, such as a hotel ticket that can be bought by anyone, then if the user locks a hotel ticket and does nothing, then no one else can buy that hotel ticket. Pushing the cross-shard operations to clients means that we also rely on the client to be honest to guarantee liveness. In 2PC-style protocols (i.e. S-BAC), the liveness property relies on the shard being honest instead.