error-groups: tx validation, execution and txpool error codes

[s1na]

This is my attempt at refactoring error codes from #650, and based on the conversation in #817. The core of it is re-categorising "execution" and "gas" error groups. I am proposing separation by "tx-validation" and "runtime" errors. This allows the client to realize eth_sendTransaction failed and tx will not be propagated due to validation issue. A tx failing in runtime will be still included in the block.

Detail of changes and their rationale are:

• Remove standard json-rpc errors. They are defined in the upstream spec, we shouldn't need to re-define them.
• On the non-standard errors as specified in EIP-1474: I have kept -32000 and -32001 which are used in clients already. Let's add others on a need-basis.
• ZK execution errors: seems like it should be standardised between L2s?
• Re-categorise EVM related codes. New group for tx validation errors. The miner will not include a tx failing on any of these codes. New group for EVM runtime errors. That's revert 3, and the generic VM error -32015. Unfortunate they are not in a range.
• Changed eth_simulateV1 to use the groups properly. One of the things we realized is that the validation errors and the "simulate-specific" codes cross-cut in the number range.

The "non-standard" codes -32002/-32003 are used as well. But unfortunately with different meanings: geth: timeout / response-too-large; reth/nethermind: transaction-rejected / resource-unavailable. This needs discussing.