when priorities are equal do breadth first search

[Eh2406]

Now that #298 has landed, the next steps are evaluate additional heuristics on a case-by-case basis. The obvious next most important from #291 is discovery_order. Thinking about it, almost all users could use this as a tiebreaker, and it costs next to nothing to calculate and the storage cost in priorities seems minimal. Overall, the cost to adding it to the documentation and telling all of our users to add it to their boilerplate seems higher than just always including it in the priority.

Given that we would now have a reasonable default I thought about giving prioritize a default implementation. Unfortunately, it would need to return a Self::Priority and associated types cannot have a default value. :-(

cargo run -r -- -m pub --with-solana --filter solana-archiver-lib -t 16 is now giving 2.52min, a massive improvement from 7.56min with dev.

[codspeed-hq]

CodSpeed Performance Report

Merging #299 will degrade performances by 6.94%

_{Comparing Eh2406:discovery_order (bda2b81) with dev (d49bf5f)}

Summary

❌ 1 regressions
✅ 5 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	dev	Eh2406:discovery_order	Change
❌	sudoku-hard	4 ms	4.2 ms	-6.94%