[Future] Move away from documenting "null safety" as its own entity

[parlough]

Currently we have a lot of documentation covering null safety as both an integrated and an isolated concept, but it really is just a part of the type system. In the longer term (far post Dart 3) "null safety" as a term likely shouldn't be super common on the site, but instead be incorporated naturally across the entire documentation, particularly updated type and type systems sections.

Prerequisites:

• The "Language" umbrella #4588
• [Sound null safety] Language feature docs #4584

Some steps:

• Consider a new nullable type section or subsection to new language docs
• Verify nullable and non-nullable types are covered and used as necessary across all docs on the site
• Introduce specific or better examples in cheatsheets and/or codelabs around nullable types
• Update or rewrite type system article to be more thorough in nullable and non-nullable types (Understanding null safety has great related content)

Way later steps:

• Remove all mentions of unsound null safety (--no-sound-null-safety)
• Remove or archive articles about enabling and configuring null safety

[MaryaBelanger]

Small subset of this, but there's been conversation about ?. and getting the useful info from https://dart.dev/null-safety/understanding-null-safety#smarter-null-aware-methods into the "core docs", specifically the operators page where ?. is "discussed" (there's like, a sentence about it at the bottom of the page).

From @fishythefish :

Do we have a good reference page for what ?. does? I've found https://dart.dev/language/operators#other-operators and https://dart.dev/language/classes#using-class-members, but I'm wondering if there are any others. I've noticed two common misunderstandings of this operator in google3:

1. Some code uses ?. for every property access. I can sort of see the reasoning prior to null safety, but many of these uses flow into code that's going to blow up on null anyway. It feels like users think ?. magically avoids NPEs altogether.
2. A lot of code (and some reviewers) assume that once you have ?. somewhere in a property chain, you need ?. on every subsequent access in the chain. This one's understandable - if you don't know about the short-circuiting behavior, it's reasonable to read foo?.bar.baz as (foo?.bar).baz and then "correct" it to foo?.bar?.baz. I didn't find any documentation of this behavior, even in the spec.

And also links to this: https://github.com/dart-lang/language/blob/main/accepted/2.12/nnbd/feature-specification.md#null-aware-operator

Maybe this should be handled in a break out issue, just documenting here since it's part of the null safety break down.

[fishythefish]

A clarification from @srawlins in response to point 2:

That's also reasonable because there was no null-shortening before null safety.

I didn't realize this was something added in null safety until I went back and looked at the feature specification again. Analyzer does emit a diagnostic for unnecessary uses of ?. so it's fairly easy to clean up those uses, but I've had a few code reviewers confused by this change in boilerplate.