Support property accesses on ~Escapable types

[natecook1000]

This adds support for having nonescapable types on the LHS of property accesses in the #expect(...) and #require(...) macros. This includes some additional changes that could lead to supporting a nonescapable result type in #require(...), but that may need more lifetime support in the language.

Motivation:

In the BinaryParsing library, I have a nonescapable ParserSpan type. In the tests, I need to write #expect(span.count == 0) instead of the more fluent #expect(span.isEmpty).

Modifications:

I've enabled lifetime annotations in the Testing and test targets, and then added overloads of __checkPropertyAccess and __checkValue that allow ~Escapable parameters.

Checklist:

• Code and documentation should follow the style of the Style Guide.
• If public symbols are renamed or modified, DocC references should be updated.

[stmontgomery]

Quick note up front that currently we expect our main branch to remain buildable using a 6.1 toolchain. (No further back than that, luckily.) So you may need #if compiler(>=6.2) guards in some places.

[grynspan]

I asked Nate to look at branching from #840 instead of main here as the implementations are different and we wouldn't want to accidentally introduce something on main that was unsupportable under the new implementation.