phd: include error cause chain in failure output

[hawkw]

Currently, when phd tests fail by returning an anyhow::Error, the code for formatting the error message only includes the top-level error. When an error has a complex cause chain --- as most of the internal framework errors in phd do, due to use of anyhow::Context --- this means that most of the error information is discarded in the test output. For example, a test failure might currently look like this:

2024-01-16T21:07:12.693405Z  INFO phd_runner::execute: 117: test phd_tests::migrate::can_migrate_from_head ... FAILED: Failed to spawn migration source VM
    error backtrace: disabled backtrace

failures:
    phd_tests::migrate::can_migrate_from_head

test result: FAILED. 0 passed; 1 failed; 0 skipped; 0 not run; finished in 3.36s

This branch changes the error output to use the default fmt::Debug implementation for anyhow::Error, which includes the top-level error, the entire cause chain if the error has a cause, and the backtrace, if one was captured.

Now, the output for the same test failing looks like this:

2024-01-16T21:07:55.858780Z  INFO phd_runner::execute: 117: test phd_tests::migrate::can_migrate_from_head ... FAILED: Failed to spawn migration source VM

Caused by:
    0: constructing test VM
    1: building environment for new VM
    2: setting up VM execution environment
    3: Failed to extract file 'propolis-server' from tarball '/tmp/phd-propolis/ab0f89332b9ef2e21a1f4b41401b9d944ea2e6159f7c0c85143dd1c0f7e0c338/propolis-server.tar.gz'
    4: No file named 'propolis-server' found in tarball

failures:
    phd_tests::migrate::can_migrate_from_head

test result: FAILED. 0 passed; 1 failed; 0 skipped; 0 not run; finished in 3.36s

Or, if RUST_BACKTRACE=1 is set, like this:

2024-01-16T21:13:25.663321Z  INFO phd_runner::execute: 117: test phd_tests::migrate::can_migrate_from_head ... FAILED: Failed to spawn migration source VM

Caused by:
    0: constructing test VM
    1: building environment for new VM
    2: setting up VM execution environment
    3: Failed to extract file 'propolis-server' from tarball '/tmp/phd-propolis/ab0f89332b9ef2e21a1f4b41401b9d944ea2e6159f7c0c85143dd1c0f7e0c338/propolis-server.tar.gz'
    4: No file named 'propolis-server' found in tarball

Stack backtrace:
   0: anyhow::__private::format_err
             at /home/eliza/.cargo/registry/src/index.crates.io-6f17d22bba15001f/anyhow-1.0.75/src/lib.rs:678:13
   1: phd_framework::artifacts::store::extract_tarball
             at ./framework/src/artifacts/store.rs:537:9
   2: phd_framework::artifacts::store::extract_tar_gz::{{closure}}
             at ./framework/src/artifacts/store.rs:487:20
   3: phd_framework::artifacts::store::extract_tar_gz
             at ./framework/src/artifacts/store.rs:479:5

    < .. BACKTRACE SNIPPED FOR BREVITY ... >

  28: std::rt::lang_start
             at /rustc/82e1608dfa6e0b5569232559e3d385fea5a93112/library/std/src/rt.rs:166:17
  29: main
  30: _start_crt
  31: _start

failures:
    phd_tests::migrate::can_migrate_from_head

test result: FAILED. 0 passed; 1 failed; 0 skipped; 0 not run; finished in 3.99s

The additional information from the error's cause chain should come in handy while debugging test failures.