Raise ToolError and exit non-zero on ecosystem panics

Summary
--

The huge number of changes in
#22205 (comment) should have
obviously been a red flag, but I think it would be nice if CI failed when new
ecosystem panics were introduced. This PR adds a check for diagnostic lines that
start with `panic: Panicked at crates/`, raises a `ToolError` if any are found
in the results from the comparison executable, and then also exits non-zero if
any errors are returned.

If exiting non-zero is going too far, we could also just raise the `ToolError`,
as that will at least trigger this message, which was not the case on the
PLR1712 PR:

https://github.com/astral-sh/ruff/blob/f14edd8661e2803254f89265548c7487f47a09f6/python/ruff-ecosystem/ruff_ecosystem/check.py#L103-L106

Another option would be not to exit zero if Ruff panics, even if `--exit-zero`
is used, but I saw that ty has the same behavior and assumed that that was
intentional.

Test Plan
--

Local testing on the 0.15.3 tag showing that ruff-ecosystem exited non-zero. I
can also introduce a panic into a lint rule to test this in CI

Author: ntBre

python/ruff-ecosystem/ruff_ecosystem/check.py

@@ -48,6 +48,8 @@
     r"^(?P<diff>[+-])? ?(?P<location>.*): (?P<code>[A-Z]{1,4}[0-9]{3,4}|[a-z\-]+:)(?P<fixable> \[\*\])? (?P<message>.*)"
 )
 
+PANIC_DIAGNOSTIC_LINE_RE = re.compile(r"^[^:]+: panic: Panicked at crates/")
+
 CHECK_VIOLATION_FIX_INDICATOR = " [*]"
 
 GITHUB_MAX_COMMENT_LENGTH = 65536  # characters
@@ -529,6 +531,10 @@ async def compare_check(
         comparison_task.result(),
     )
 
+    for line in comparison_output:
+        if PANIC_DIAGNOSTIC_LINE_RE.match(line):
+            raise ToolError(line)
+
     diff = Diff.from_pair(baseline_output, comparison_output)
 
     return Comparison(diff=diff, repo=cloned_repo)

python/ruff-ecosystem/ruff_ecosystem/cli.py

@@ -105,10 +105,13 @@ def entrypoint():
         for signal in [SIGINT, SIGTERM]:
             loop.add_signal_handler(signal, main_task.cancel)
         try:
-            loop.run_until_complete(main_task)
+            had_errors = loop.run_until_complete(main_task)
         finally:
             loop.close()
 
+        if had_errors:
+            exit(1)
+
 
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(

python/ruff-ecosystem/ruff_ecosystem/main.py

@@ -96,7 +96,7 @@ async def limited_parallelism(coroutine: Awaitable[T]) -> T:
         case _:
             raise ValueError(f"Unknown output format {format}")
 
-    return None
+    return bool(result.errored)
 
 
 async def clone_and_compare(