Remove bottleneck in proof generation of Merkle map

[jpraynaud]

Content

This PR includes fix of some bottlenecks identified in the proof generation of the Merkle map:

• Fix bottleneck when Merkle map computes a proofs.
• Fix bottleneck when computing the list of certified items in a Mekle map proof.
• Refactor Merkle map benchmarks to make them more accurate.

Benchmarks

Total Requests	Concurrency	Transactions/Request	Pool (x50): Request/s	Pool (x50) + Optimization MKMap proof computation: Request/s
1000	100	1	281.4	1403.97
1000	100	2	192.84	468.39
1000	100	3	138.47	289.79
1000	100	4	123.5	241.31
1000	100	5	102.83	149.93
1000	100	6	84.58	137.74
1000	100	7	73.35	111.12
1000	100	8	65.93	99.63
1000	100	9	60.79	80.79
1000	100	10	56.39	79.55

Running on Linux / 8 cores / 64 GB RAM / 2 TB SSD

Pre-submit checklist

• Branch
  • [x] Tests are provided (if possible)
  • [x] Crates versions are updated (if relevant)
  • [x] CHANGELOG file is updated (if relevant)
  • [x] Commit sequence broadly makes sense
  • [x] Key commits have useful messages
• PR
  • [x] No clippy warnings in the CI
  • [x] Self-reviewed the diff
  • [x] Useful pull request description
  • [x] Reviewer requested

Issue(s)

Closes #1730

[github-actions[bot]]

Test Results

    3 files  ±0     43 suites  ±0   8m 32s :stopwatch: -14s 1 029 tests +1  1 029 :white_check_mark: +1  0 :zzz: ±0  0 :x: ±0  1 127 runs  +1  1 127 :white_check_mark: +1  0 :zzz: ±0  0 :x: ±0 

Results for commit 4f85efb1. ± Comparison against base commit bda0888f.

:recycle: This comment has been updated with latest results.