Fix UnicodeDecodeError on non-UTF-8 log lines

Read the access log in binary mode so decoding happens per-line inside
the existing contextlib.suppress block. Previously readlines() decoded
the whole file as UTF-8 up front and threw on any raw non-UTF-8 byte
(e.g. a scanner's malformed URL) before truncate(0) could run, so the
bad line was never cleared and every subsequent scrape failed.

app.py

@@ -21,8 +21,8 @@ async def read_and_process_logs() -> dict[tuple[str, str], dict[str, float]]:
     metrics_data: dict[tuple[str, str], dict[str, float]] = {}
 
     if os.path.exists(unit_log_file):
-        async with aiofiles.open(unit_log_file, mode="r+") as f:
+        async with aiofiles.open(unit_log_file, mode="rb+") as f:
-            lines: list[str] = await f.readlines()
+            lines: list[bytes] = await f.readlines()
             await f.truncate(0)
 
         for line in lines: