Claude review: rust: update `kernel::c_str!` documentation

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Summary:** Updates the doc comment on `c_str!` to explain that `c"literal"` syntax should be preferred for string literals, and that `c_str!` remains useful for non-literal compile-time strings (e.g. `concat!(file!(), ...)`). Also adds a comment explaining why a `compile_error!` arm for literal arguments was intentionally omitted.

**Positive observations:**

1. The updated doc comment at line 379-383 is clear and well-written, correctly explaining the relationship between `c"..."` literals and the macro.

2. The new "Panics" section (lines 385-387) documenting the interior NUL panic behavior is a good addition — the old documentation only said "should not contain any `NUL` bytes" without specifying what happens.

3. The example is well-chosen:
   ```rust
   // This is allowed, but `c"literal"` should be preferred for literals.
   const BAD: &CStr = c_str!("literal");
   
   // `c_str!` is still needed for static non-literal C strings.
   const GOOD: &CStr = c_str!(concat!(file!(), ":", line!(), ": My CStr!"));
   ```
   Naming the first example `BAD` and the second `GOOD` makes the intent immediately obvious.

4. The NB comment at lines 402-404 explaining the design decision not to add a `compile_error!` arm for `$str:lit` is valuable — it preempts the natural follow-up question of "why not just reject literals?" and explains that it would break macro authors who happen to have a literal at the top of expansions.

**Minor observations:**

- The "Panics" section says the macro "panics" — technically it's a compile-time `panic!` (since it's used in a `const` context), which is a compilation error, not a runtime panic. This is arguably a pedantic distinction and the current wording matches Rust documentation conventions, so it's fine as-is.

**Verdict:** Looks good. No issues found.

---
Generated by Claude Code Patch Reviewer