Handle JoinSet task errors in test utility functions

[coderabbitai]

Problem

In src/tool/subcommands/api_cmd/test_snapshot.rs and src/tool/subcommands/api_cmd/generate_test_snapshot.rs, background tasks that consume JoinSet results currently drop them silently using the pattern:

tokio::spawn(async move { while services.join_next().await.is_some() {} });

This ignores Result variants for panics and cancellations returned by join_next(). The standard pattern used elsewhere in the codebase (e.g., chain_sync/, daemon/mod.rs) is to check results with while let Some(res) and handle errors appropriately.

Context

A similar issue was fixed in src/tool/offline_server/server.rs in PR #6493, but attempts to fix it in the test utility files caused codecov CI to constantly timeout, requiring the changes to be reverted.

Files to Fix

• src/tool/subcommands/api_cmd/test_snapshot.rs
• src/tool/subcommands/api_cmd/generate_test_snapshot.rs

Related

• PR: test: unit test for start_rpc #6493
• Comment: test: unit test for start_rpc #6493 (comment)
• Reported by: @hanabi1224

[0xDevNinja]

Hi! Picking this up. Plan: replace the inline &mut JoinSet::new() in ctx() of both test_snapshot.rs and generate_test_snapshot.rs with a named JoinSet whose draining is moved into a detached tokio::spawn that logs errors via tracing::warn! (matching the while let Some(res) pattern in daemon/mod.rs:propagate_error and chain_sync/tipset_syncer.rs). Detaching avoids the codecov-timeout issue hit by earlier inline-await attempts.

Will open a PR referencing this issue via Closes #6505 shortly. Happy to be assigned if appropriate, otherwise the PR linkage should suffice.

— @0xDevNinja