use GFT survey questions to define flags

[damonmcc]

resolves #814

changes

• add seed question_flags.csv to declare the 1-to-1 relationship between the GFT survey questions and the GFT survey flags
  • lot zoning is the only exception because it isn't a binary flag, it's a category
• revise variables.csv to declare the many-to-many relationship between the survey flags and the source data
• change build logic to produce a flag column and an ID column for each survey question

todos

• [ ] update intermediate models to align with new final columns so that outputs are correct (e.g. different buffers on the same dataset depending on the flag)

notes

in the new question_flags seed:

• term "field name" is used because our output columns are FGDB fields that Jack will be choosing and aliasing
• each flag has a value (Yes/No) and there are IDs of geometries that determined the flag's value. so I ended up with _flag and _flag_id. maybe _flag_ids would be better though
• was tempted to lean more into survey_question or the survey_question_id values that Jack made for the app, but he cautioned that those IDs could change even if the essence of the question doesn't. so seemed good to just declare flag_field_name values that are 1-to-1 with survey questions

will do after this

• update sources__ to align with new survey flags and reflect the latest requests from the GIS team

tests

Successful build here, which includes all dbt tests passing

a row from test_actual__pilot_projects

[damonmcc]

while there are rows in variables.csv that aren't 1-to-1 with a survey question, these two tests will fail