Computation of Merkle proof has bottleneck with multiple transactions

[jpraynaud]

Why

We have noticed that the computation of the proof done by the Merkle map is not optimal and leads to a performance drop when multiple transactions are certified by the prover.

What

Fix the bottleneck which creates a performance drop when the number of transactions to certify increases.

How

• [x] Investigate the performance drop
• [x] Fix the performance drop
• [x] Benchmark the performances improvements