Downstairs reports: database disk image is malformed

Core(s) generated while omicron was running with 187 disks and one instance. The instance had two disks attached.

Many core files were generated from omicron, with the 2nd oldest being this one from inside crucible.

Core file on catacomb at:
/data/staff/core/gimlet-sn21/20221015/core.oxz_propolis-server_1b5ebd28-3a8f-4013-8927-99c200dbd501.propolis-server.24772

$ mdb core.oxz_propolis-server_1b5ebd28-3a8f-4013-8927-99c200dbd501.propolis-server.24772 
Loading modules: [ libumem.so.1 libc.so.1 ld.so.1 ]
> $C ! demangle
fffff5ffd07fe470 libc.so.1`_lwp_kill+0xa()
fffff5ffd07fe4a0 libc.so.1`raise+0x1e(6)
fffff5ffd07fe4f0 libc.so.1`abort+0x58()
fffff5ffd07fe500 panic_abort::__rust_start_panic::abort::h5e4ff9722552943b+9()
fffff5ffd07fe510 __rust_start_panic+9()
fffff5ffd07fe5c0 rust_panic+0x1c()
fffff5ffd07fe670 std::panicking::rust_panic_with_hook::ha80c5cb4a11249f7+0x1fe()
fffff5ffd07fe6c0 std::panicking::begin_panic_handler::{{closure}}::h1c465a251fe8dc5f+0xc5()
fffff5ffd07fe6f0 std::sys_common::backtrace::__rust_end_short_backtrace::h73c8afa9397a6a09+0x20()
fffff5ffd07fe730 rust_begin_unwind+0x71()
fffff5ffd07fe770 core::panicking::panic_fmt::h476086f4291cd29f+0x33()
fffff5ffd07fe850 core::panicking::assert_failed_inner::h1c6c74356de69ab3+0xca()
fffff5ffd07fe8b0 core::panicking::assert_failed::ha8afc05e959f3059+0x46()
fffff5ffd07feb00 crucible::Downstairs::retire_check::h3b5d34ae9aea2742+0x4dd()
fffff5ffd07fed90 tokio::loom::std::unsafe_cell::UnsafeCell<T>::with_mut::hf1fc0701748dd1a8+0xaa4()
fffff5ffd07ffb10 tokio::runtime::task::harness::Harness<T,S>::poll::ha036dcbe5e688841+0x58()
fffff5ffd07ffbc0 std::thread::local::LocalKey<T>::with::hfaf80dd10aedc955+0x1c8()
fffff5ffd07ffc40 tokio::runtime::scheduler::multi_thread::worker::Context::run_task::h07f63c15f77994f4+0x21f()
fffff5ffd07ffd00 tokio::runtime::scheduler::multi_thread::worker::Context::run::ha897fdb5c3963887+0x72e()
fffff5ffd07ffd50 tokio::runtime::scheduler::multi_thread::worker::run::h2ec47ea0d8bd33fe+0x75()
fffff5ffd07ffd90 tokio::runtime::task::core::CoreStage<T>::poll::h0b439faa5b8bdc5a+0x3f()
fffff5ffd07ffe10 tokio::runtime::task::harness::Harness<T,S>::poll::hbf631161bda3ca94+0x49()
fffff5ffd07ffec0 tokio::runtime::blocking::pool::Inner::run::h24bd9cdb22619ff2+0xbf()
fffff5ffd07fff10 std::sys_common::backtrace::__rust_begin_short_backtrace::hee5d124e97635faa+0x109()
fffff5ffd07fff60 core::ops::function::FnOnce::call_once{{vtable.shim}}::h2fcf5432bb9edc09+0x78()
fffff5ffd07fffb0 std::sys::unix::thread::Thread::new::thread_start::h87fc76126371d726+0x27()
fffff5ffd07fffe0 libc.so.1`_thrp_setup+0x6c(fffff5ffee75a240)
fffff5ffd07ffff0 libc.so.1`_lwp_start()

Start by inspecting the reported core dump and the crucible::Downstairs::retire_check stack frame, using the core file path in the issue. Determine why the malformed database disk image reaches this assertion and reproduce the failure; done means the crash is prevented and the malformed-image case is covered by an appropriate regression test.