ui: Ensure dc selector correctly shows the currently selected dc

[johncowen]

This PR solves one thing, but its one of those things that affects other things:

1. I noticed that our DC selector had stopped updating depending on certain things including non existing DCs and the local datacenter etc.
2. I'd left a test FIXME in here to come back and resolve that I wasn't sure of initially, turns out if was all related.
3. When I fixed this, the DOM order of the (what seems like) a completely unrelated test changed. Our problematic scrollpane's DOM does not always equal what you see visually. The test was testing the DOM order, but we already had a step that we write from ages ago that asserts the human visual order rather than the DOM order, which is what we should probably be asserting anyway. Without the addition of vertically the test was asserting the DOM order (theres some whitespace there also which explains what looks like extra change)

[evrowe]

Is the primary fix for this issue then opting away from using the get helper (which IIRC can introduce some caching problems)? Seems like a sound approach, just wanting to make sure I understand the intent

[johncowen]

The primary fix here was undo-ing the mistaken "return a function" signature of the cached-model helper which you can see up there ^ somewhere. When I originally did it in the PR where it went in I vaguely remember playing around with different approaches. Typically somehow ended up with half of one approach and half of another.

For the get helper usage I realized I could re-wangle the code to just use the let that we are already using. Here get was being used in multiple places to access a property on an object returned by another helper. So I just passed the resulting object in through the let instead and accessed the property on that object as normal (see below where we access dc.Name instead of just dc). Previously dc was a string, now it's an object with a string Name prop.

[evrowe]

Yeah, I could see what was happening with using the value returned by the let helper, and the resulting changes to dc's type, I was just interested to see whether the get helper itself was the problem or if the changes here were follow-on to addressing the root cause elsewhere.

Speaking of, appreciate the explanation re: cached-model, that helps clear things up for me :)

[hc-github-team-consul-core]

🍒 If backport labels were added before merging, cherry-picking will start automatically.

To retroactively trigger a backport after merging, add backport labels and re-run https://circleci.com/gh/hashicorp/consul/485541.