Missing command from failed context.run_shell_command invocations

[schneems]

Context

When using run_shell_command in an integration test:

    TestRunner::default().build(
        config,
        |context| {
            println!("{}", context.pack_stdout);
            assert_contains!(context.pack_stdout, "# Heroku Ruby Buildpack");
            assert_contains!(
                context.pack_stdout,
                r#"`BUNDLE_BIN="/layers/heroku_ruby/gems/bin" BUNDLE_CLEAN="1" BUNDLE_DEPLOYMENT="1" BUNDLE_GEMFILE="/workspace/Gemfile" BUNDLE_PATH="/layers/heroku_ruby/gems" BUNDLE_WITHOUT="development:test" bundle install`"#);

            assert_contains!(context.pack_stdout, "Installing puma");

            let secret_key_base = context.run_shell_command("echo \"${SECRET_KEY_BASE:?No SECRET_KEY_BASE set}").stdout;

Expected

If the command fails the full command will be presented to the user on error.

Actual

The command is not always presented back to the user, in this case here's the output:

thread 'test_default_app_latest_distro' panicked at /Users/rschneeman/.cargo/registry/src/index.crates.io-6f17d22bba15001f/libcnb-test-0.25.0/src/test_context.rs:194:43:
Error running container:

docker command failed with exit code 2!

## stderr:

bash: -c: line 1: unexpected EOF while looking for matching `"'

## stdout:



note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    test_default_app_latest_distro

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 6 filtered out; finished in 13.85s

It will show for other cases like trying to run "doesnotexist" but it would be easier to see that there's a missing quotation mark if the full command is visible in the error output like:

Error running command `echo "${SECRET_KEY_BASE:?No SECRET_KEY_BASE set}`

...

Or something similar.