As a small quality of life improvement for debugging, I think we should crate a newtype wrapper around the VM's ExecutionError that overrides the Display impl:

#[derive(Debug, thiserror::Error)]
pub struct ExecError(pub ExecutionError);

impl Display for ExecError {
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        f.write_str(&PrintDiagnostic::new(&self.0).to_string())
    }
}

This allows us to always use anyhow::Result<()> in tests instead of a mix between that and miette::Result. The latter is only needed because ExecutionError does not turn into a nice report by default. E.g. if we do tx_context.execute_code(code).await? in a test with anyhow::Result, we'd get:

one or more warnings were emitted
Error: assertion failed at clock cycle 1889 with error message: get_balance can only be called on a fungible asset

Using ExecError + anyhow::Result instead, we get:

stderr ───
  one or more warnings were emitted
  Error:   x assertion failed at clock cycle 1889 with error message: get_balance can only be called on a fungible asset
      ,-[~/git/miden-base/target/test-dev/build/miden-lib-f0afd6e1413cd388/out/asm/kernels/transaction/lib/asset_vault.masm:62:5]
    61 |     dup exec.account_id::is_fungible_faucet
    62 |     assert.err=ERR_VAULT_GET_BALANCE_CAN_ONLY_BE_CALLED_ON_FUNGIBLE_ASSET
      :     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    63 |     # => [faucet_id_prefix, faucet_id_suffix, vault_root_ptr]
      `----

So the task here is to:

• Introduce the above error type in crates/miden-testing/src/tx_context/errors.rs.
• Change TransactionContext::execute_code to return Result<ExecutionOutput, ExecError>.
• Replace all miette::Result in miden-testing with anyhow::Result. This means we also need to get rid of a bunch of into_diagnostic calls.
• Make adjustments to the assert_execution_error! macro to handle matching on ExecError(error).