Separate the lottery evaluation and trial selection + other improvements

[skim725]

Addresses #73.

[shippy]

Either delete this line (as it merely prints one possible selection to stdout) or save it into a variable and use it in the for loop.

[shippy]

No way around it: this changes API for pickAndEvaluateTrial in a way that would break UVRA. Since the function is never invoked, it doesn't actually need to - pickAndEvaluateTrial can be labeled deprecated, and its innards mostly factored out into showLotteryEvaluation (which should perhaps be showTrialEvaluation?).

Prior to merge:

• revert argument change (actually, all changes) in pickAndEvaluateTrial (UVRA depends on being able to pass DataObject to it)
• replace eligible logic in pickAndEvaluateTrial (which is almost everything past line 13) with showLotteryEvaluation
• cut out unnecessary code
• document the introduced functions better

[shippy]

Sorry, just noticed this — this does not describe the function at all. (Or rather, it describes lines 23 and 24 of the function.)

[shippy]

This will not work on Matlab versions prior to 2016b be — don't ask me why, but Matlab was very strict that strings are denoted by apostrophes, not by quotes.

Also, note that this code doesn't create a basecase for a block whose name is neither monetary nor medical, which means that it will fail hard in those cases.

[shippy]

The way to generalize that for tasks with different kinds of conditions would be to a) poll fieldnames in block.conditions and make them into table headers, b) populate the table with the condition assignments extracted from each block. This is perfect for a vectorized approach -- in fact, cellfun(@(block) struct2table(block.conditions), blocks) would be a one-line expression of that logic (haven't checked).