[ui, deployments] Promote Canary and Unhealthy Allocations in the deployment status panel

[philrenaud]

This PR upgrades the importance of the "Promote Canary" functionality and adopts conventions around unhealthy allocations during a deployment.

A big side-effect here is the realization that allocation.deploymentStatus.Healthy isn't always just null or true; it can be false as well. This adds an explicit unhealthy status into the UI for when an allocation is running but fails its health checks.

Screenshots

Canary health being checked:

Canaries are running, but some have failed their health checks:

Some canaries have gone into a failed clientState:

Canaries have passed their health checks, require manual promotion:

Resolves #17525

[github-actions]

Ember Test Audit comparison

	main	26db164	change
passes	1475	1473	-2
failures	0	2	+2
flaky	0	0	0
duration	11m 18s 517ms	000ms	-11m 18s 517ms

[philrenaud]

This is the little top-right-corner status indicator that shows up when enough of an alloc exist to summarize it in a "+N" block

[philrenaud]

This is the indicator for when an allocation is un-aggregated/summarized, in say a small enough cluster that each allocation gets its own square.

[philrenaud]

deploymentIsAutoPromoted is a new property for the UI, and slightly changes our messaging during the "checking your allocs' health" phase.

[philrenaud]

humanize to drop the _ in health_unknown.

(in case you were wondering: we don't use unknown here because that has a dedicated orange colour associated on elements from "Node status unknown" / max client disconnect stuff.)

[philrenaud]

I'm not thrilled that I'm nesting a ternary here but it is only causing me mild discomfort

[philrenaud]

This is the change to the <legend> beneath the allocations themselves.