Guarded byte string and C string non-raw literals should not report "unknown prefix"

Code

macro_rules! a {
    ($($t:tt)*) => {};
}

a! {
  b#""#
  c#""#
}

Current output

Compiling playground v0.0.1 (/playground)
error: prefix `b` is unknown
 --> src/lib.rs:6:3
  |
6 |   b#""#
  |   ^ unknown prefix
  |
  = note: prefixed identifiers and literals are reserved since Rust 2021
help: consider inserting whitespace here
  |
6 |   b #""#
  |    +

error: invalid string literal
 --> src/lib.rs:6:4
  |
6 |   b#""#
  |    ^^^^
  |
  = note: unprefixed guarded string literals are reserved for future use since Rust 2024
help: consider inserting whitespace here
  |
6 |   b# ""#
  |     +

error: prefix `c` is unknown
 --> src/lib.rs:7:3
  |
7 |   c#""#
  |   ^ unknown prefix
  |
  = note: prefixed identifiers and literals are reserved since Rust 2021
help: consider inserting whitespace here
  |
7 |   c #""#
  |    +

error: invalid string literal
 --> src/lib.rs:7:4
  |
7 |   c#""#
  |    ^^^^
  |
  = note: unprefixed guarded string literals are reserved for future use since Rust 2024
help: consider inserting whitespace here
  |
7 |   c# ""#
  |     +

error: could not compile `playground` (lib) due to 4 previous errors

Desired output

Compiling playground v0.0.1 (/playground)
error: invalid string literal
 --> src/lib.rs:6:4
  |
6 |   b#""#
  |   ^^  ^
  |
  = note: guarded byte string literals are not supported
help: did you mean to write a raw byte string literal?
  |
6 |   br#""#
  |    +

error: invalid string literal
 --> src/lib.rs:7:4
  |
7 |   c#""#
  |   ^^  ^
  |
  = note: guarded C string literals are not supported
help: did you mean to write a raw C string literal?
  |
7 |   cr#""#
  |    +

error: could not compile `playground` (lib) due to 2 previous errors

Rationale and extra context

The error message prefix `b` is unknown is incorrect. The prefix b for string literals does mean something, just not for guarded string literals.

Rust Version

1.88.0-nightly (2025-04-02 d5b4c2e4f19b6d703737)

(sorry this isn't the rustc --version --verbose output asked; i've tested with the online playground only)

Anything else?

Cursor::c_or_byte_string should lex a # character and produce a different error token than UnknownPrefix. It currently doesn't.

Reproduce the issue with the Rust snippet, then read Cursor::c_or_byte_string in compiler/rustc_lexer/src/lib.rs around the referenced line. Trace how the # character is handled for b and c literals; done means the diagnostics match the desired output without separate unknown-prefix errors.