fix(core): avoid panic when workflow channels close

[sachinsharma3191]

Summary

• avoid panics in workflow processing when fetch/activation channels are already dropped
• stop workflow processing loop gracefully in that case and log instead of panicking
• add focused regression tests for channel-forwarding behavior

Test plan

• cargo test -p temporalio-sdk-core forwarding_output_stops_when_activation_channel_dropped -- --nocapture
• cargo test -p temporalio-sdk-core forwarding_output_succeeds_with_live_channels -- --nocapture
• cargo test -p temporalio-sdk-core core_tests::workers::worker_can_shutdown_after_never_polling_ok -- --nocapture
• cargo test -p temporalio-sdk-core core_tests::workers::shutdown -- --nocapture
• cargo fmt --all --check

[chris-olszewski]

We do not want to ignore cases where receivers get dropped unexpectedly. This is unexpected behavior and should result in a panic.

Were you encountering these panics in your own code? If so, please open an issue.

Ah, I see the branch references #692. I think we can solve this more gracefully than just ignoring unexpected receiver drops.Ah, I see the branch references #692. I think we can solve this more gracefully than just ignoring unexpected receiver drops.