Switch from Buffalo Pop to Gorm

[jacobconley]

In my time familiarizing myself with Pop, I've noticed a few things that make it not as great a fit for us as I first imagined:

• One of the original advantages was its integration with the rest of the Buffalo ecosystem - we aren't using as much of it as I thought we would
• Doesn't support things like composite primary key (pretty model-centric), foreign keys are awkward, weird UUID default
• The documentation isn't stellar
• SQLite support was an issue, so we had to do this farcical postgresql containerization for unit tests and stuff. I'm sure there's another workaround but it was inconvenient
• The configuration and Genny options and what not are alright, but there were some behaviors that we don't want and would have to code around (again due to its intended goal of being a part of the Buffalo ecosystem)

The Buffalo project is still great and worthy of support in its capacity as a quick all-in-one Rails-style MVC web framework in Golang, but these attributes conflict with our goal of providing a general unopinionated framework that extends beyond that domain.

Notwithstanding the above, Gorm has these advantages:

• Strong independent support with thorough configuration, documentation, and testing
• Powerful features like DryRun mode, Upserts and locks
• Kinda funny name

I'm not yet sure how migrations will compare to Soda, but I'm already convinced enough that switching to Gorm is a good idea

[jacobconley]

Closed due to the disintegration in da0306c; #24 will remain open for further notes and discussion