Bad error snippet for cargotest failure

Start with the linked pull-request comment and reproduce the snippet produced for the cargotest failure. Trace the log-analyzer entry point that selects excerpts, then compare its output with the failure block shown in this issue. Done means the snippet includes the failing test and its actual error instead of unrelated build and test-suite sections.

I am not sure how the bot decided what to put into the snippet here but it doesn't include the actual problem. What I would have hoped for is something like this:

failures:

---- alt_registry::no_api stdout ----
running `/Users/runner/work/rust/rust/build/aarch64-apple-darwin/stage2-tools/aarch64-apple-darwin/release/cargo check`
running `/Users/runner/work/rust/rust/build/aarch64-apple-darwin/stage2-tools/aarch64-apple-darwin/release/cargo login --registry alternative`

thread 'alt_registry::no_api' (2024432) panicked at src/tools/cargo/tests/testsuite/alt_registry.rs:821:10:

test failed running `/Users/runner/work/rust/rust/build/aarch64-apple-darwin/stage2-tools/aarch64-apple-darwin/release/cargo login --registry alternative`
error: could not exec process `/Users/runner/work/rust/rust/build/aarch64-apple-darwin/stage2-tools/aarch64-apple-darwin/release/cargo login --registry alternative`: could not execute process `/Users/runner/work/rust/rust/build/aarch64-apple-darwin/stage2-tools/aarch64-apple-darwin/release/cargo login --registry alternative` (never executed)

Caused by:
    Broken pipe (os error 32)

Stack backtrace:
   0: <std::backtrace::Backtrace>::create
   1: <std::io::error::Error as anyhow::context::ext::StdError>::ext_context::<cargo_util::process_error::ProcessError>
   2: <cargo_util::process_builder::ProcessBuilder>::output
   3: <cargo_util::process_builder::ProcessBuilder>::exec_with_output
   4: <cargo_test_support::Execs>::run
   5: testsuite::alt_registry::no_api
   6: <testsuite::alt_registry::no_api::{closure#0} as core::ops::function::FnOnce<()>>::call_once
   7: test::__rust_begin_short_backtrace::<core::result::Result<(), alloc::string::String>, fn() -> core::result::Result<(), alloc::string::String>>
   8: test::run_test::{closure#0}
   9: std::sys::backtrace::__rust_begin_short_backtrace::<test::run_test::{closure#1}, ()>
  10: <std::thread::lifecycle::spawn_unchecked<test::run_test::{closure#1}, ()>::{closure#1} as core::ops::function::FnOnce<()>>::call_once::{shim:vtable#0}
  11: <std::sys::thread::unix::Thread>::new::thread_start
  12: __pthread_cond_wait
stack backtrace:
   0: __rustc::rust_begin_unwind
   1: core::panicking::panic_fmt
   2: cargo_test_support::panic_error::pe
   3: <cargo_test_support::Execs>::run
   4: testsuite::alt_registry::no_api
   5: <testsuite::alt_registry::no_api::{closure#0} as core::ops::function::FnOnce<()>>::call_once
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
---- alt_registry::no_api stdout end ----

failures:
    alt_registry::no_api

test result: FAILED. 3771 passed; 1 failed; 391 ignored; 0 measured; 0 filtered out; finished in 581.98s

Instead, it shows random parts of the LLVM build, the rustdoc test suite, and ends with some random test suite in the middle of the log where nothing failed.