content: Update to reflect changes to template system in v0.146.0

[jmooring]

This will affect many pages (e.g., everywhere we show an example path such as layouts/_default/single.html, directory tree structures, etc.), so we may want to start with just the lookup order page to reduce merge conflicts with the changes for #2996.

Need to review:

• Reimplement and simplify Hugo's template system hugo#13541
• Add new standard layout 'all' (in addition to single,list) hugo#13545
• Deprecate the _internal/some-template construct hugo#13553

Also:

• Revise inline partial examples too: https://gohugo.io/templates/partial/#inline-partials
• Need to change URLs to embedded templates (shortcodes, render hooks, etc.). See data/embedded_template_urls.toml. This is addressed in #2992.
• Check override instructions (the template path) for alias.html, robots.txt, rss.xml, sitemap.xml, and sitemapindex.xml.
• Document partial naming. See this comment, which is a bit confusing: tpl: Fix language handling in partials hugo#13614 (comment)

[jmooring]

I'm not convinced we have to document this (the previous behavior was undocumented/lucky), but this no longer works (and that's fine):

{{- with .Params.blocks }}
  <aside class="content-aside">
    {{- range . }}
      {{- $t := printf "partials/blocks/%s.html" . }}
      {{- if (templates.Exists $t) }}
        {{- partial $t $ }}
      {{- end }}
    {{- end }}
  </aside>
{{- end }}

In the above we're doing this:

{{ partial "partials/foo.html" }}

We throw an ignorable warning about this, and error if the partial template cannot be found.

[jmooring]

@bep Please confirm. Going forward, when defining and calling inline partials, one should do this:

{{ partial "foo.html" }}

{{ define "_partials/foo.html" }}
   FOO
{{ end }}

instead of this:

{{ partial "foo" }}

{{ define "partials/foo" }}
   FOO
{{ end }}

[bep]

{{ define "_partials/foo.html" }}

Both currently works, but I think we should use the _partials prefix for both inline and file partials going forward, at least for consistency. We eventually need to deprecate the old syntax so we can remove the "double checks".