* Only show the top 40 blog tags
In https://github.com/pulumi/pulumi-hugo/pull/215, I had suggested
that instead of physically deleting tags we didn't want to show, we
compute it algorithmically, by only showing the "Top N" tags. This
commit introduces said functionality.
This has a few advantages:
* Preserves old metadata (the authors added the tags because they
felt they were meaningful and captured information about the posts).
* Enables us to surface those tags differently in the future (who
knows, maybe someday we'll want to run a "spinnaker" campaign).
* Notably, also keeps the tag index pages, which Google has indexed.
* Enables us to add a "View More ..." link at the bottom of the
page if folks want to see the entire list.
* Perhaps most importantly, protects against future bloat. For
example, since this tag cleanup happened, we have added top-level
tags for "aliases", "app-runner", "iam", "open-source", and
"refactoring", each of which has only a single post.
I chose 40 as the N in Top N, because that's how many we show today.
I could see an argument for filtering this based on post count
instead (e.g., only those with >3 posts).
* Add back some tags
Now that we filter out unpopular tags, we can add back some of the
ones previously removed.
* Eliminate "real languages" from our website
We know that this terminology is unfriendly and can be seen as
gatekeeping. We've moved away from it over time, but there are many
traces of it in our documentation -- most notably, even on our
homepage. Instead of "real languages", let's use "familiar languages",
and in places where deeper explanation makes sense in context, we
can say "familiar, general-purpose languages."
The current 2nd-level blurb on the homepage isn't very good right now
anyhow, so this is a convenient change. As we make progress on various
marketing efforts, we will likely make more extensive changes, but
for now this at least fixes this problem. We recently A/B tested
updates without much of an impact, so let's move ahead with this change
and then if we want to test alternatives to "familiar", we can do that.
For instance, we considered "popular", etc.
* Also eliminate "real code"
