perf: improve metric performance

[twoeths]

Motivation

• gauge.inc() takes 2.45% cpu time which is > 2 / 3 runHeartbeat() cpu time

Description

• gauge.inc() will hash the label objects and push to Map every time we call it
• so we we write some cached gauge metric class to only do that when scrape
  • gauge has no label: use NoLabelCachedGauge, this gives 7x improvement
  • gauge has 1 label: use OneLabelCachedGauge, this gives 1.2x improvement
  • gauge has 2 labels: use TwoLabelCachedGauge, this gives 2.5x improvement
• also remove unnecessary gauge metric

[dapplion]

also remove unnecessary gauge metric

@tuyennhv Could you do this item in a separate PR?

About the optimization, is it best to do it here or on Lodestar?

[codecov-commenter]

Codecov Report

Merging #287 (43db65b) into master (475c861) will decrease coverage by 1.50%.
The diff coverage is 21.06%.

❗ Current head 43db65b differs from pull request most recent head dc3c677. Consider uploading reports for the commit dc3c677 to get more accurate results

@@            Coverage Diff             @@
##           master     #287      +/-   ##
==========================================
- Coverage   80.46%   78.96%   -1.51%     
==========================================
  Files          42       42              
  Lines        9102     9384     +282     
  Branches      826      825       -1     
==========================================
+ Hits         7324     7410      +86     
- Misses       1778     1974     +196     

Impacted Files	Coverage Δ
src/metrics.ts	25.20% <21.02%> (+2.11%)	⬆️
src/index.ts	69.42% <25.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 475c861...dc3c677. Read the comment docs.

[twoeths]

closing as this PR was merged #433