[PWGLF] Updates on a structure of resonanceModuleInitializer

[mjkim525]

Summary

This PR fixes an incorrect daughter-table grouping and the resulting CPU-time regression in resonanceModuleInitializer.

Previously, the modular initializer used the reduced ResoCollision as the grouping key, while Track, V0, and Cascade candidates still referenced the original aod::Collision. As a result, the original collision association was not available to the framework grouping mechanism. This could cause the full filtered Track table to be processed for every reduced collision, producing incorrect cross-collision associations and unnecessary CPU overhead.

Changes

• Preserve the existing two-stage module-initializer architecture.
• Add a association between each reduced ResoCollision and its original aod::Collision.
• Use the original collision as the grouping key in the daughter initializer.
• Retrieve the corresponding reduced collision through a SmallGroups association.
• Apply automatic collision grouping to Track, V0, and Cascade candidates.

[github-actions]

O2 linter results: ❌ 0 errors, ⚠️ 28 warnings, 🔕 2 disabled

[alibuild]

Error while checking build/O2Physics/code-check for 5060d6e at 2026-08-24 16:05:

## sw/BUILD/O2Physics-latest/log
c++: fatal error: Killed signal terminated program cc1plus
c++: fatal error: Killed signal terminated program cc1plus
ninja: build stopped: subcommand failed.

Full log here.