feat: ETH RPC: Use Block Cache for EthGetBlockByHash

[aarshkshah1992]

Closes #10520.

Part of #12293.

[aarshkshah1992]

Tested this on a mainnet node

With Caching enabled (fetch all txdata as well)

First request

time curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"eth_getBlockByHash","params":["0xd5742c7ead47c4224dd1e7a877532fc52fdc3094bad36b85b66661b2c2cf2401", true],"id":1}' http://localhost:1234/rpc/v1

real	0m0.060s

Second request

time curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"eth_getBlockByHash","params":["0xd5742c7ead47c4224dd1e7a877532fc52fdc3094bad36b85b66661b2c2cf2401", true],"id":1}' http://localhost:1234/rpc/v1

real	0m0.034s

[akaladarshi]

LGTM.
Just a couple of comments.

[Stebalien]

Do you have any before metrics?

[aarshkshah1992]

@Stebalien Lemme get them for you.

[Stebalien]

Seems like a reasonable approach.

[aarshkshah1992]

@Stebalien

Without any caching

1st fetch

time curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"eth_getBlockByHash","params":["0xd5742c7ead47c4224dd1e7a877532fc52fdc3094bad36b85b66661b2c2cf2401", true],"id":1}' http://localhost:1234/rpc/v1

real	0m0.073s

2nd fetch

time curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"eth_getBlockByHash","params":["0xd5742c7ead47c4224dd1e7a877532fc52fdc3094bad36b85b66661b2c2cf2401", true],"id":1}' http://localhost:1234/rpc/v1
real	0m0.065s

Caching definitely helps here. Keep in mind that this is a snapshot synced node with splitstore pruning enabled. Difference should be even more pronounced on an archival node with no splitstore.

[Stebalien]

2x savings actually aren't that large here... I wonder if that's just because the network is quiet right now.

On the other hand, the cache is pretty small so it probably isn't harmful.

[rvagg]

This seems fine but I am a little concerned about the large memory footprint of lotus just getting bigger, turn on the eth API and you have yet another blob in memory. Although thankfully it only grows if you use it.
I'm not sure exactly how much these take up but the JSON output of one of your example blocks is over 500Kb, so maybe they end up being something like 200Kb in memory, so perhaps 100Mb here when we cache 500 of them.