Refactor acquire

[matteolimberto-da]

This is a slight refactoring of the lifecycle code, where I tried to remove some complexity from the process.

Specifically, the acquire function would

• project a claim
• evaluate observables and calculate acquisition time of sub-contracts
• consume Cond nodes and fix acquisition time of stochastic Whens
• collect the Either and the sub-contracts' acquisition time
• embed the claim ( which is then projected again by the subsequent functions

Following the refactoring, acquire' takes care only of steps 2 and 3 (the "business logic"), which in my opinion simplifies it. It also maintains clear separation between

• the acquisition time of contract c
• the acquisition time of any sub-contract of c

The exercise, lifecycle and expire functions take as input an "acquired claim" (that is, a claim where we know if we acquire its sub-contracts and at which time).

The change puts us in a position to separate the acquire into an effectful part and a non-effectful one. I am planning to use the latter as part of the valuation semantics (to propagate a potentially-stochastic acquisition time).

Happy to hear your feedback.

[matteolimberto-da]

Please, note that this change make the exercise function stop the recursion when an election is made (same change as in https://github.com/digital-asset/contingent-claims/pull/74) in order to avoid the bug described in the linked PR. The same considerations apply.

[matteolimberto-da]

Something I forgot to mention in the main comment is that before this refactoring we were forced to consume the Cond node in the acquire' function (for the reason explained in https://github.com/digital-asset/contingent-claims/pull/65).

Now, this is no longer the case and acquire' does not necessarily have to remove the Cond node. By not consuming Cond we avoid creating an additional version of the instrument on the ledger. This might be the object of a subsequent PR.