Identity PII

### Clear and concise description of the problem

Some languages e.g., Rust, make printing models, headers, etc. very easy e.g.,

```rust
#[derive(Debug)]
pub struct Employee {
pub id: Option,
pub name: Option,
pub title: Option,
}

let employee = client.get_employee(&id).await?;
println!("Found {employee:#?}");

// Found Employee {
// id: Some("abc123"),
// name: Some("Mary Smith"),
// title: Some("Director"),
// }
```

Devs idiomatically use this for logging or printf-debugging but `id` and probably `name` shouldn't be logged as PII (note: does not affect display in debuggers).

What we'd like is a way in TypeSpec to identity which fields either have PII (opt-in) or wouldn't reasonably have PII (opt-out) so that we can implement or derive this `Debug` trait so that we don't emit those fields e.g.,

```typespec
model Employee {
@pii
id?: string,
name: string,
title?: string,
```

In which case, our emitter could emit something like:

```rust
#[derive(SafeDebug)]
pub struct Employee {
#[safe_debug(hidden)]
pub id: Option,
pub name: Option,
pub title: Option,
}

let employee = client.get_employee(&id).await?;
println!("Found {employee:#?}");

// Found Employee {
// name: Some("Mary Smith"),
// title: Some("Director"),
// ..
// }
```

Other languages might also elide this information from any tracing they might do, or `ToString()` override, etc.

### Checklist

- [x] Follow our [Code of Conduct](https://github.com/azure/typespec-azure/blob/main/CODE_OF_CONDUCT.md)
- [x] Check that this issue is about the Azure libraries for typespec. For feature request in the typespec language or core libraries file it in the [TypeSpec repo](https://github.com/Microsoft/TypeSpec/issues/new/choose)
- [x] Read the [docs](https://azure.github.io/typespec-azure/).
- [x] Check that there isn't already an issue that request the same feature to avoid creating a duplicate.

No files or tests are named. Start by locating the TypeSpec Azure emitter that generates models and determine how a PII annotation could flow into generated language-specific debugging or string representations; done should define the annotation semantics and demonstrate that marked fields are omitted from generated logging output.

Written by the indexing model from the issue text.