fix empty_node_cache use for cache

[silentEAG]

Try to solve #6577
update empty_node_cache promptly.

[ngundotra]

ty for posting this, this will save a bunch of CU on mainnet!

How did you find?

[silentEAG]

Hey @ngundotra, thanks for the confirmation. I am an auditor from Sec3. I noticed this when I was looking at the concurrency Merkle Tree for something else. Apologies if submitting a PR is not the best way to bring up such issues.

Basically, I found that the empty_node_cached is used directly without utilizing empty_node_cache in four places.

Two of them (in the initialize function) could use the cache and have already been addressed by a previous PR.

The remaining two instances are:

• concurrent_merkle_tree.rs#L172: This does not involve loop calculations, so it doesn't need to use the cache and can be ignored.

• concurrent_merkle_tree.rs#L277: This involves loop calculations and could potentially benefit from calculating the cache first before iterating. However, I think the overlaps are relatively small in most cases. The performance gain may not be that significant. Maybe this could also be ignored.

My last commit was trying to make cargo clippy happy. Thanks again for reviewing my changes!