perf: remove allocations from merkle tree proof verification logic

[kevaundray]

Issue Addressed

This PR does two small things:

• Removes the allocations that were happening on each loop
• Makes it more explicit that the bit in the index is only being used to specify the order of the inputs for the hash function

Additional Info

Benchmarked on Apple M4 Max and there is a 33% performance increase, which concretely amounted to ~1.2 microseconds savings for depth=32.
Not a lot, so I think the main performance benefit would be the reduction in possible allocations that would have happened during the for loop.

[michaelsproul]

Looks great, nice improvement thanks!

[mergify]

Merge Queue Status

✅ The pull request has been merged at 3b97964

This pull request spent 37 minutes 31 seconds in the queue, including 36 minutes 9 seconds running CI.
The checks were run on draft #8621.

Required conditions to merge

• check-success=local-testnet-success
• check-success=test-suite-success