Optimize client batching with set operations

[yuqian19870725-creator]

Summary

• return client IDs with character history as a set
• split bubble recipients using set intersection and difference instead of repeated list membership scans
• add regression coverage for the set return type

This changes the recipient partitioning step from repeated linear membership checks to set operations, avoiding quadratic behavior for large fleet battles.

Validation

• Python syntax compilation passed for the implementation and test files
• focused microbenchmark for recipient partitioning:
  • 1,000 clients: 3.025 ms to 0.018 ms (about 172x faster)
  • 5,000 clients: 74.280 ms to 0.177 ms (about 419x faster)
• these figures cover only the recipient partitioning operation, not the complete server tick
• the full repository test requires the unavailable blue runtime module in this environment