feat: support dialog behaviours

[jlp-craigmorten]

See https://developer.mozilla.org/en-US/docs/Web/Accessibility/ARIA/Roles/dialog_role#possible_effects_on_user_agents_and_assistive_technology

When the dialog is correctly labeled and focus is moved to an element (often an interactive element, such as a button) inside the dialog, screen readers should announce the dialog's accessible role, name and optionally description, along with announcing the focused element.

Although this is not normative, but MDN's opinion (with a disclaimer saying as such), testing with NVDA and VoiceOver this appears to be the case and seems not unreasonable to implement.

Originally posted by @jlp-craigmorten in #12 (comment)

[cmorten]

Have commented in w3c/aria#1195 (comment) where the associated PR w3c/aria#2054 adds clarity that this will likely be a SHOULD expectation on assistive technology.

As the draft stands:

Assistive technology SHOULD give precedence to exposing author defined dialog accessible descriptions when a dialog is invoked and user focus is moved to a descendant of the dialog element.