Reactive gas price oracle

[uprendis]

rework gasprice oracle to rely on two parts:

• Constructive part: increases gasprice according to accumulated gas power of validators (less -> higher price). This part is needed to protect network from draining validators. This part is needed because minGasPrice/baseFee balancing cannot be done very quickly per-block due to epoch-based consensus structure
• Reactive: takes gas price from top G's gas position in an average of txpool statistic for last 12 seconds, where G depends on chosen certainty (higher certainty -> smaller G). Certainty is a percentile of eth_feeHistory

Final gasprice estimation is a maximum of these 2 parts

Previously gasPrice estimation was based only on constructive part. This PR adjusts weights for the constructive part and improves calculation of accumulated gas power of validators

Note: like before, meeting the constructive part is not optional but mandatory, it serves as a soft minGasPrice/baseFee limit. This PR adds ftm_effectiveBaseFee API call for retrieving current soft limit for gas price

[hadv]

LGTM

[hadv]

nit: I think it'd better to rename this method to PendingTxs

Suggested change

	func (pool *TxPool) PendingSlice() types.Transactions {
	func (pool *TxPool) PendingTxs() types.Transactions {

[uprendis]

Oh it's just because there's already a method called Pending which returns a map, I wanted to emphasise the difference between them

[rus-alex]

go test ./gossip failed:
*dummyTxPool does not implement TxPool (missing PendingSlice method)

[cyberbono3]

I suggest to increase test coverage of gossip/gasprice package
ok github.com/Fantom-foundation/go-opera/gossip/gasprice 0.148s coverage: 0.0% of statements [no tests to run]

[uprendis]

@cyberbono3 added tests in e4ddb6d

[BMItr]

👌 great work