Semester simulation framework

[bmiszalski]

• Planning and development of Semester class
• Alteration of existing classes to fit Semester class logic

[bmiszalski]

Best approach is to keep current framework (mostly) and do the block acceptance/management via updating a clone of the science database (with nvisits=1 for all blocks and perhaps other things simplified, e.g. no time critical, etc.). A good approach here is to have a WipeBlocks function in the Queue class that takes all the blocks that were scheduled in a queue and writes those as done in the database. Some may be chosen at random to be rejected (e.g. seeing went too high).

We also want to fundamentally show how the impact of the SA's choice of object-by-object scheduling compares against a nightly generated schedule. This can be done in practice using scores in the database (Paul): to simulate a night scheduled by an SA, we would pick the highest scored block and observe it. Then we choose the next highest scored block at the end of the observation and go to that. There should also be a random factor here that chooses to go for a realignment instead of a block (and that realignment lasts an amount of time between say 15-25 min), or some other telescope fault, to simulate real observing.