When we introduced Markdown support, we added a single rich_text field

for each supported model, and all fields were tied to that setting.

However, if a person wanted to use Markdown for one field but not

another, they were out of luck. That also required that we perform some

complex escaping/unescaping of all fields on a model when switching text

modes.
This is the first change in a series that splits the rich_text field,

providing per-field flags. The changes are all inter-dependent, but

split up based on layers of the codebase.
This change adds the new rich_text fields, updates logic and templates

to point to the new fields, and removes usage of the old fields.

This testing required the other related changes:

Unit tests pass
Editing fields in the UI set those fields to be in Markdown format, but

  left other fields un-touched.
Fields with a rich_text=false render as plain text, but render as Markdown

  with rich_text=true.