Update the correct column if HIV HSI event didn't run

[tamuri]

This was a nasty bug to track down!

While profiling, I noticed the population dataframe had grown from 494 to 567 columns during the simulation. New columns were being added as the simulation ran. The column names were property, which is a built-in Python keyword. No hint as to where it was coming from. Each of these columns contained a single False for one person, all the rest nan. After a lot of laborious stepping through code, traced issue to changes introduced in #1616

property is a Python keyword (most commonly used as the @property decorator). The variable property set in apply() is not included in never_ran(). As a result, each time never_ran() is called, a new column is created.

I've made following changes:

• Renamed the variable to prep_column to avoid shadowing the built-in property
• Moved the column name logic into a function so it can be used in both apply() and never_ran()

[tbhallett]

Wow, did not expect that to happen. Thanks for finding and fixing.

[tdm32]

Thanks Asif for spotting and fixing that