Update Redshift json implementation to use SUPER data type

[bcodell]

Per comment below from @matthieu-carmeille the project would benefit redshift users to leverage the SUPER data type in redshift. Likely implementation needs:

• Cast feature json column as super in build_json macro
• Ensure attribute extraction when building dataset uses super-compatible extraction syntax

Original message:
Thank you so much @bcodell for the quick response. Next we should look into the column type: it is currently character varying, it would be way more performant with SUPER type in Redshift.

Originally posted by @matthieu-carmeille in #75 (comment)

[bcodell]

@matthieu-carmeille spun up a new issue. Does this scope seem about right? And is this something you could take on? I can tackle myself if you're too busy, though in that case I could use your help with testing as I don't have access to a redshift cluster of my own. Let me know - thanks!

[matthieu-carmeille]

Hi @bcodell , please find here the related PR