feat(txpool): Nonce cache

[calbera]

similar, but simpler impl of #350

[itsdevbear]

@davidterpay just for ref

[codecov[bot]]

Codecov Report

Merging #571 (ad16e90) into main (5bf5173) will decrease coverage by 0.69%. The diff coverage is 8.98%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #571      +/-   ##
==========================================
- Coverage   63.79%   63.11%   -0.69%     
==========================================
  Files         100      100              
  Lines        4259     4316      +57     
==========================================
+ Hits         2717     2724       +7     
- Misses       1444     1494      +50     
  Partials       98       98              

Impacted Files	Coverage Δ
cosmos/x/evm/plugins/txpool/mempool/mempool.go	0.00% <0.00%> (ø)
cosmos/x/evm/plugins/txpool/plugin.go	0.00% <0.00%> (ø)
eth/core/chain_reader.go	0.00% <0.00%> (ø)
eth/core/chain_subscriber.go	0.00% <0.00%> (ø)
eth/core/state/statedb.go	56.25% <ø> (+1.14%)	:arrow_up:
cosmos/x/evm/plugins/state/plugin.go	74.18% <50.00%> (-0.82%)	:arrow_down:
cosmos/x/evm/keeper/host.go	100.00% <100.00%> (ø)
eth/core/state/journal/suicides.go	100.00% <100.00%> (ø)

[calbera]

current version is the logic used by geth’s light client txpool. now have to support pending and queued txs from geths core txpool

[itsdevbear]

Light client doesn't support queued

[calbera]

Also adding support for the .Error check, I think we should do this (as Geth does) whenever accessing anything that could have caused an error (nonce, balance specifically)

[calbera]

Also we are now exposing the state plugin to be read from multiple threads, so do we need to lock there?

[calbera]

This PR should just introduce read-only changes, namely enabling RPC methods: GetPoolNonce, SubscribeNewTxsEvent

[itsdevbear]

@corduroybera can u make ticket bls

[calbera]

So I don’t think we need to merge this one. Implementing it via geth’s txpool provides more features and more testing, we just have to configure it correctly

[itsdevbear]

I still think that approach is hood, but if you can find a way to make it seamless I'm on board.

If anything we should build a wrapper that lives in the eth folder.

[calbera]

I was originally doing a wrapper in eth/core/ but I realized there’s no need for it. It works without the wrapper. Either you have the geth TxPool in the blockchain object or you have the plug-in import it directly. The former involves ugly duplication of state

[calbera]

Here its in the blockchain: https://github.com/berachain/polaris/pull/580/files#diff-732b1d3ebcc7a1ce9c153310d8f40fbd45cbc54b72627b3d298727e6643cf318

Here is the (unused) wrapper PolarisTxPool: https://github.com/berachain/polaris/pull/580/files#diff-f2bae00423d965dd5d83cb38ea73ba24ca6560730a3b1abc828ffa9eec2f441b