perf: only compute start offset for overlong lines (#5811)

Moves the computation of the `start_offset` for overlong lines to just
before the result is returned. There is a slight overhead for overlong
lines (double the work for the first `limit` characters).

In practice this results in a speedup on the CPython codebase. Most
lines are not overlong, or are not enforced because the line ends with a
URL, or does not contain whitespace. Nonetheless, the 0.3% of overlong
lines are a lot compared to other violations.

### Before
![selected
before](https://github.com/astral-sh/ruff/assets/9756388/d32047df-7fd2-4ae8-8333-1a3679ce000f)
_Selected W505 and E501_

![all
before](https://github.com/astral-sh/ruff/assets/9756388/98495118-c474-46ff-873c-fb58a78cfe15)
_All rules_

### After
![selected
after](https://github.com/astral-sh/ruff/assets/9756388/e4bd7f10-ff7e-4d52-8267-27cace8c5471)
_Selected W505 and E501_

![all
after](https://github.com/astral-sh/ruff/assets/9756388/573bdbe2-c64f-4f22-9659-c68726ff52c0)
_All rules_

CPython line statistics:
- Number of Python lines: 867.696
- Number of overlong lines: 2.963 (0.3%)

<details>

Benchmark selected:
```shell
cargo build --release && hyperfine --warmup 10 --min-runs 50 \                                                  
  "./target/release/ruff ./crates/ruff/resources/test/cpython/ --no-cache -e --select W505,E501"
```

Benchmark all:
```shell
cargo build --release && hyperfine --warmup 10 --min-runs 50 \                                                  
  "./target/release/ruff ./crates/ruff/resources/test/cpython/ --no-cache -e --select ALL"
```

Overlong lines in CPython

```shell
cargo run -p ruff_cli -- check crates/ruff/resources/test/cpython/Lib --no-cache --select=E501,W505 --statistics
```

Total Python lines:
```shell
find crates/ruff/resources/test/cpython/ -name '*.py' | xargs wc -l
```

</details>

(Performance tested on Mac M1)

CONTRIBUTING.md

@@ -419,6 +419,13 @@ Summary
   159.43 ± 2.48 times faster than 'pycodestyle crates/ruff/resources/test/cpython'
 ```
 
+To benchmark a subset of rules, e.g. `LineTooLong` and `DocLineTooLong`:
+
+```shell
+cargo build --release && hyperfine --warmup 10 \
+  "./target/release/ruff ./crates/ruff/resources/test/cpython/ --no-cache -e --select W505,E501"
+```
+
 You can run `poetry install` from `./scripts/benchmarks` to create a working environment for the
 above. All reported benchmarks were computed using the versions specified by
 `./scripts/benchmarks/pyproject.toml` on Python 3.11.

crates/ruff/src/rules/pycodestyle/helpers.rs

@@ -53,16 +53,13 @@ pub(super) fn is_overlong(
     task_tags: &[String],
     tab_size: TabSize,
 ) -> Option<Overlong> {
-    let mut start_offset = line.start();
+    // Each character is between 1-4 bytes. If the number of bytes is smaller than the limit, it cannot be overlong.
+    if line.len() < limit.get() {
+        return None;
+    }
+
     let mut width = LineWidth::new(tab_size);
-
-    for c in line.chars() {
-        if width < limit {
-            start_offset += c.text_len();
-        }
-        width = width.add_char(c);
-    }
-
+    width = width.add_str(line.as_str());
     if width <= limit {
         return None;
     }
@@ -91,6 +88,17 @@ pub(super) fn is_overlong(
         }
     }
 
+    // Obtain the start offset of the part of the line that exceeds the limit
+    let mut start_offset = line.start();
+    let mut start_width = LineWidth::new(tab_size);
+    for c in line.chars() {
+        if start_width < limit {
+            start_offset += c.text_len();
+            start_width = start_width.add_char(c);
+        } else {
+            break;
+        }
+    }
     Some(Overlong {
         range: TextRange::new(start_offset, line.end()),
         width: width.get(),