Fix bare except clause in alembic/util/messaging.py

[koteshyelamati]

Replace bare except: with except Exception: in the status() context manager.

The status() context manager prints a status message and a completion/failure indicator. Its bare except: catches BaseException, meaning KeyboardInterrupt or SystemExit during the operation would trigger the failure message and then re-raise — but the intent is clearly to catch operational exceptions, not signals. except Exception: is the correct type here.

[MohammedAlkindi]

Ran this on Windows 11 (10.0.26200), CPython 3.13.13, against main at c116cbc0. Failing-test set is identical to base (13 tests, the test_post_write and ScriptNamingTest timezone clusters, both failing on main here) — no regressions, none fixed.

One thing worth deciding deliberately before this lands, because it is a behaviour change rather than a pure lint fix. The bare except: is inside a context manager that re-raises immediately:

@contextmanager
def status(status_msg, newline=False, quiet=False):
    msg(status_msg + " ...", newline, flush=True, quiet=quiet)
    try:
        yield
    except:
        if not quiet:
            write_outstream(sys.stdout, "  FAILED\n")
        raise

Because it re-raises unconditionally, this is one of the few genuinely defensible bare excepts — it swallows nothing. What it catches is everything, deliberately, so that the status line gets closed out. Narrowing to except Exception: means KeyboardInterrupt and SystemExit no longer print FAILED, so a Ctrl-C during a migration now leaves the status line dangling at Running upgrade ... with no terminator, where today it prints FAILED.

That may well be what you want — arguably Ctrl-C is not a failure and shouldn't be labelled one. But it is a user-visible change to interrupt handling, not a no-op cleanup, so it seemed worth surfacing rather than letting it ride in as an E722 fix.

If the intent is purely to satisfy the linter while keeping the behaviour, except BaseException: does that exactly. If the intent is that interrupts should stop being reported as failures, then this patch is right as written and might deserve a line in the changelog, since anyone scripting against that output would see it change.

Either way the one-line diff is correct in isolation and I have no objection to it landing — I just did not want the interrupt case to go unnoticed.