perf: partitions from actix to tokio

[craigmayhew]

Describe the changes

Migrate partition mining from Actix actors to Tokio services; extend broadcaster to support Tokio subscribers.

• Motivation

  • Simplify concurrency model, reduce Actix arbiters, unify with other Tokio services, improve testability.

• Key changes

  • New modules/types: partition_mining_service.rs, PartitionMiningService, PartitionMiningServiceInner, PartitionMiningController.
  • Broadcaster additions: MiningBroadcastEvent, SubscribeTokio, UnsubscribeTokio (best-effort).
  • Removal: addresses.rs and ActorAddresses.
  • Macro/import cleanup: $crate::storage::ie in macros; removed stray ie imports.
  • Observability: added warnings/errors when packing channel try_send is Full/Closed to surface backpressure and failures.

• Behavioral changes

  • Partition mining now runs as Tokio tasks, controlled via PartitionMiningController.
  • BroadcastMiningService fans out to both Actix and Tokio subscribers; closed Tokio channels are pruned on send failure. Unbounded channels are used.

• Chain initialization changes

  • Removed ActorAddresses and the set of PartitionMiningActor addresses.
  • Added partition_controllers: Vec<PartitionMiningController> on IrysNodeCtx.
  • Replaced init_partition_mining_actor with init_partition_mining_services returning (Vec<PartitionMiningController>, Vec<TokioServiceHandle>).
  • Call sites switched from sending MiningControl to ctrl.set_mining(should_mine).

• Operational considerations

  • Packing channel now logs when saturated/closed, aiding visibility into backpressure.
  • BroadcastMiningService should be registered before miners spawn (tests ensure via from_registry()).

• Testing

  • Tests migrated to tokio::test where appropriate and ensure broadcaster init before miners.
  • Added/updated comments in epoch snapshot tests to clarify partition expiration and repacking behavior.

Follow up PR / out of scope

• Replace the Actix broadcaster with a Tokio-native bus and remove all actix references from partition_mining_service.rs.
• Remove Actix System/Arbiter usages from chain.
• Migrate non-partition tests and services off Actix patterns (e.g., #[actix::test], actix_rt::time::sleep, actix::Message usage in unrelated services).
• Remove packing sender as a param of ServiceSendersInner::init_with_sender() and move it to be a "regular" member of ServiceSenders.

Checklist

• Tests have been added/updated for the changes.
• Documentation has been updated for the changes (if applicable).
• The code follows Rust's style guidelines.

Additional Context

This is the second of three PRS for actix->tokio
Previous in series of actix removal PRs: #829
Next in series of actix removal PRs: #869

[craigmayhew]

Flagging flaky tests:

FLAKY 2/2 block_production::block_production::heavy_test_always_build_on_max_difficulty_block
FLAKY 2/2 irys-chain::mod::multi_node::mempool_tests::slow_heavy_mempool_publish_fork_recovery_test::case_1_full_validation
FLAKY 2/2 irys-chain::mod::multi_node::validation::heavy_block_invalid_evm_block_reward_gets_rejected
FLAKY 2/2 irys-p2p::tests::integration::heavy_should_not_resend_recently_seen_data
TRY 2 TMT irys-actors::data_sync_test::slow_test_data_sync_with_different_peer_performance

[craigmayhew]

Flagging flaky tests:
FLAKY 2/2 [ 12.331s] irys-chain::mod synchronization::heavy_should_resume_from_the_same_block
TRY 2 FAIL [ 11.811s] irys-chain::mod multi_node::reth_restart::reth_restarts_use_block_index_before_sync

[craigmayhew]

This has received some review via slack. Merging and will take any feedback into the next PR in the series to remove actix.