This change promotes the explanation of what set_color actually does from the very last note to the description.

I recently ran into the roughly the same misunderstanding described in #2378. Like the reporter of that issue, I also believed that set_color "set some internal shell state" and had much more magical behavior than it actually does. The changes to the documentation that stemmed from that report didn't help me avoid this misunderstanding—it was quite buried—so this is my attempt to help the next reader.

When making this change, I tried to honor some of the constraints from the original discussion, especially "keep[ing] the high-level 'it changes colors' as the first thing the user reads." I tried to lead with the user's intent, then follow it with how set_color satisfies the intent.

TODOs:

• (n/a) If addressing an issue, a commit message mentions Fixes issue #<issue-number>
• (n/a) Changes to fish usage are reflected in user documentation/manpages.
• (n/a) Tests have been added for regressions fixed
• User-visible changes noted in CHANGELOG.rst