Add targets branching

[matthew-phelps]

Branching of the targets pipeline enables two features:

• Allows user to add/modify endpoints without re-running previously added endpoints (via branch caching).
• Allows user-defined endpoints to be run in parallel. This is not enabled by this merge request as the parallel compute configuration has not been added

Regarding the endpoints (a.k.a. branch) caching, there are several caveats that will trigger re-runs of already computed endpoints:

• If the user re-orders the endpoints in the mk_endpoints_definition function, any endpoints that have changed position will be re-run
• Changes to mk_adam, criteria or statistical functions will trigger re-runs of the endpoints that use those functions (this is a feature, as it ensures the endpoints always reflect the current state of the code)