Canonical URLS are a google invention, not a W3C one, so to find out about it, you’d have to look at Google pages for developers.
So only for pages who are URLS leading to the same content, and only the ones you consider “copies” (so the original/main/indexable page does not need a canonical meta). You also don’t need the whole URL with schema and domain: they can be relative, so
<link rel=“canonical” href=“/signup-stepone” />
would be fine.