Consolidate interactive input/output capabilities into GlobalCommandOptions

## Summary

Refactor the interactivity model to expose explicit `SupportsInteractiveInput` and `SupportsInteractiveOutput` properties in a single location, rather than splitting the logic across `GlobalCommandOptions.NoPrompt` and `terminal.IsTerminal()`.

## Motivation

azd currently uses two separate mechanisms for interactivity:

- **`NoPrompt`** (in `GlobalCommandOptions`): Suppresses interactive input prompts
- **`isTerminal`** (from `terminal.IsTerminal()`): Controls output rendering (spinners, colors, terminal resize)

These two concepts are spread across different parts of the codebase with no unified model. This makes it harder to reason about what capabilities are available in a given context and creates subtle bugs when they interact.

Consolidating them into explicit properties would:

- **Make the interactivity model self-documenting** — one place to check what's supported
- **Allow commands to check capabilities uniformly** — instead of mixing `NoPrompt` checks with `isTerminal` checks
- **Simplify future enhancements** — e.g., MCP tools that can handle prompts (via external prompt endpoint) but don't support spinners
- **Reduce the cognitive load** for new contributors understanding the codebase

## Implementation Notes

- Add `SupportsInteractiveInput` and `SupportsInteractiveOutput` to `GlobalCommandOptions` or a new `InteractivityContext` type
- Derive values from: explicit flags (`--no-prompt`, `--non-interactive`), env vars, agent detection, TTY detection
- Update `Console`/`AskerConsole` to consume these properties instead of separate `noPrompt` + `isTerminal` booleans
- This is a refactoring — no user-facing behavior change expected