feat(sidecar): ScoreCache for account_states

[thedevbirb]

Related to #393, this PR introduces a wrapper over HashMap called ScoreCache that implements a scoring system on get, insert and update operations.

The map size can be configured, with a default of 1 MiB that should hold around 15k account state entries considering also its key. For account states in particular a metrics bolt_sidecar_account_states has been added to track its usage.

ScoreCache vs HashMap/ScoreCache Insert -- size: 1000
                        time:   [16.345 µs 16.745 µs 17.213 µs]
                        change: [-4.8838% -1.1181% +2.2086%] (p = 0.54 > 0.05)
                        No change in performance detected.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high mild
ScoreCache vs HashMap/HashMap Insert -- size: 1000
                        time:   [14.296 µs 14.488 µs 14.660 µs]
                        change: [-2.9083% +0.4949% +3.8702%] (p = 0.77 > 0.05)
                        No change in performance detected.
ScoreCache vs HashMap/ScoreCache Get -- size: 1000
                        time:   [15.983 µs 16.516 µs 17.003 µs]
                        change: [-3.5447% +5.5430% +15.125%] (p = 0.22 > 0.05)
                        No change in performance detected.
ScoreCache vs HashMap/HashMap Get -- size: 1000
                        time:   [13.273 µs 13.756 µs 14.232 µs]
                        change: [-5.4725% +1.4866% +8.8014%] (p = 0.69 > 0.05)
                        No change in performance detected.
ScoreCache vs HashMap/ScoreCache Update -- size: 1000
                        time:   [16.918 µs 17.405 µs 17.815 µs]
                        change: [-9.9428% -2.4709% +5.4745%] (p = 0.54 > 0.05)
                        No change in performance detected.
ScoreCache vs HashMap/HashMap Update -- size: 1000
                        time:   [13.179 µs 13.487 µs 13.747 µs]
                        change: [-2.3176% +4.8491% +12.955%] (p = 0.22 > 0.05)
                        No change in performance detected.
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe

[thedevbirb]

Open for a first review!

[merklefruit]

Love it! Great work

[merklefruit]

Great!

[estensen]

Consider varying sizes of operations (1000, 10k, 100k) to understand how perf varies

[thedevbirb]

Since I want to measure the raw overhead of modifying the score when doing get, insert, update operations I don't think there is a big advantage in testing different sizes. Also it slows down the benchmarking by really a lot. Just 1000 iterations takes a couple of mins on a M3 pro cpu

[estensen]

But don't wanna block this from merging

[thedevbirb]

But don't wanna block this from merging

No worries! It was indeed great feedback. Pushing a commit soon to address it

[mempirate]

Nice job, LGTM