["model"] Refactor calculations to avoid repeated methodological questions

[PhilMiller]

Summary

People keep asking about why/how the new infections, admissions, and census are calculated. We have every reason to believe the code is correct but it's hard for new readers of the code (including myself, when I started, and still) to follow. We can revise it without changing the output to clarify what it's doing.

Additional details

Suggested fix

Express the computation exactly as you would the derivation or an explanation - cf 'literate code'.

[PhilMiller]

@cjbayesian @quinn-dougherty

Is the current dispositions_df useful in itself, possibly for some future feature, or is it purely an intermediate for the calculation of admissions and census?

[PhilMiller]

Incidentally, this refactoring also makes #206 into a one-line fix: change the day 0 value of new_infections from 0.0 to i

[PhilMiller]

E.g. #222