Remove lexer-based comment range detection (#5785)

## Summary

I'm doing some unrelated profiling, and I noticed that this method is
actually measurable on the CPython benchmark -- it's > 1% of execution
time. We don't need to lex here, we already know the ranges of all
comments, so we can just do a simple binary search for overlap, which
brings the method down to 0%.

## Test Plan

`cargo test`

crates/ruff_python_ast/src/helpers.rs

@@ -717,7 +717,7 @@ pub fn map_subscript(expr: &Expr) -> &Expr {
 }
 
 /// Returns `true` if a statement or expression includes at least one comment.
-pub fn has_comments<T>(node: &T, locator: &Locator) -> bool
+pub fn has_comments<T>(node: &T, locator: &Locator, indexer: &Indexer) -> bool
 where
     T: Ranged,
 {
@@ -732,26 +732,9 @@ where
         locator.line_end(node.end())
     };
 
-    has_comments_in(TextRange::new(start, end), locator)
-}
-
-/// Returns `true` if a [`TextRange`] includes at least one comment.
-pub fn has_comments_in(range: TextRange, locator: &Locator) -> bool {
-    let source = &locator.contents()[range];
-
-    for tok in lexer::lex_starts_at(source, Mode::Module, range.start()) {
-        match tok {
-            Ok((tok, _)) => {
-                if matches!(tok, Tok::Comment(..)) {
-                    return true;
-                }
-            }
-            Err(_) => {
-                return false;
-            }
-        }
-    }
-    false
+    indexer
+        .comment_ranges()
+        .intersects(TextRange::new(start, end))
 }
 
 /// Return `true` if the body uses `locals()`, `globals()`, `vars()`, `eval()`.