Which move is the best default when built-in types and none become part of this workflow?

Use built-in types to match the real shape of the data: counts as integers, text as strings, flags as booleans, and missing optional data as None. That is the stronger engineering move because candidates who model data honestly sound more trustworthy because their reasoning starts from the problem rather than from syntax trivia. Forcing everything into strings or treating missing data as a fake value creates confusion that shows up later in comparisons and validation.