accounts-db: Use PubkeyHasher in SlotCache

[vadorovsky]

Problem

SlotCache used pubkeys as keys, yet it used the AHash hasher.

Summary of Changes

To imporove performance of lookups, use PubkeyHasher.

Fixes #

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.2%. Comparing base (c859d8b) to head (91a1a47).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7307   +/-   ##
=======================================
  Coverage    83.2%    83.2%           
=======================================
  Files         797      797           
  Lines      361114   361114           
=======================================
+ Hits       300736   300761   +25     
+ Misses      60378    60353   -25     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[brooksprumo]

To imporove performance of lookups, use PubkeyHasher.

Ooh, can you share perf results from this change?

[vadorovsky]

To imporove performance of lookups, use PubkeyHasher.

Ooh, can you share perf results from this change?

Well, I can definitely say that pubkey hasher is in general faster:

    #[test]
    fn test_hashmapz() {
        let pubkeys1: Vec<_> = (0..1_000_000).map(|_| Pubkey::new_unique()).collect();
        let pubkeys2 = pubkeys1.clone();

        let mut ahash_map: HashMap<Pubkey, (), ahash::RandomState> =
            HashMap::with_capacity_and_hasher(1_000_000, ahash::RandomState::new());
        let mut pubkeyh_map: HashMap<Pubkey, (), PubkeyHasherBuilder> =
            HashMap::with_capacity_and_hasher(1_000_000, PubkeyHasherBuilder::default());

        let (_, ahash_measure) = measure_us!(for pubkey in pubkeys1 {
            ahash_map.insert(pubkey, ());
        });
        let (_, pubkeyh_measure) = measure_us!(for pubkey in pubkeys2 {
            pubkeyh_map.insert(pubkey, ());
        });

        println!("ahash: {ahash_measure}");
        println!("pubkey hasher: {pubkeyh_measure}");
    }

running 1 test
ahash: 305600
pubkey hasher: 256180
test bank::partitioned_epoch_rewards::calculation::tests::test_hashmapz ... ok

I ran it dozens of times and the results are similar and pubkey hasher always outperforms ahash about ~15%.

Honestly I didn't catch it with profiler. I rather just thought of it when working on #6900 (where I do profiles, but the hashing itself is not the biggest bottleneck) and I started questioning the necessity of AHash in maps with pubkeys that are not really DDOS-able. The change of hasher probably contributes a tiny bit to the overall improvement.

[brooksprumo]