There are multiple sources for the term 'string', including WEBIDL,
INFRA, Unicode, and XML Schema. In our discussions with RDF-star,
we have found that there is a need for generic string definitions
for use by JSON, Schema.org, RDF, and XML-based specs which are not
consistent with DOMString/USVString.

This PR needs to be accompanied by changes to the I18N-GLOSSARY in order
to be complete. This PR is for initial discussion purposes before we
embark on those changes.

…ages

This change significantly reorganizes the section on strings in hopes
of making the definitions clearer and to reduce the overall
repetitive nature of the text.

It also moves the text about design-principles to a note at the top,
rather than burying it in the text.

The relationship of DOMString and USVString to surrogates is clarified
and pulled into one place.

- Move inline style to local.css
- fix lint-ignore on string definition
- link to webidl definitions as needed
- add a note about why you want DOMString when specifying bytes on the
  wire

- Fix links to DOMString and USVString to use WebIDL markup
- Rephrase the sentence about UTF-8's relationship to surrogate code
  points.

We have a general practice of putting the best practice text
before the explanation of the BP so that users linking into
specdev can read below the best practice rather than having
to scroll up and start the section over.

This change moves the string definition guidance ahead of all
of the text about string types.

- Change the MUSTard to recommend USVString first.
- Split the MUSTard into two BPs
- Add a missing link around one instance of `USVString`
- Add an issue that includes a link to w3ctag/design-principles#454 and
  appropriately scary sounding text.