Resolver: refactor

[stsewd]

Simplify a couple of things, and we can remove the overrides from .com after this.

• _get_canonical_project and _get_canonical_project_data are basically the same, the former tries to recursively resolve the project for cases that we don't really support, and the second just checks for the cases that we do support and returns the relationship in case of a subproject. So I just renamed _get_canonical_project_data to _get_canonical_project.
• We were passing the project slug when resolving the path (this was a residual from where we were allowing USE_SUBDOMAIN=False).
• Resolving is now split into two steps: resolving the domain, and resolving the path.
• We were using require_https for .com only, this was since on .com we were using the https attribute to track the progress of a custom domain, this is no longer the case, all custom domains on .com are https.
• _use_custom_domain is the same as _use_cname.
• Two more methods to resolve a path were added, they are basically the same as resolve, but they work on the object itself, instead of passing each part separately. This results in fewer queries in case the version object is already in memory.

[ericholscher]

Looked at this today but didn't have enough time to fully review this. Will give it a longer look tomorrow, but this looks like a solid win!

[ericholscher]

This looks like a good refactor 👍