Fix inflation calculation to account for total stake

[joepetrowski]

Currently (source), the staking rewards calculation calculates the total amount at stake by multiplying the SlotStake - the stake of the minimally staked validator - by the number of validators.

The actual amount at stake - if we consider any funds liable to be slashed as "at stake" - then the number of tokens at stake is higher since most validators will have more than the slot stake.

The calculation should find the sum of the amount at stake (sum of exposure.total from Stakers) and pass that to the payout function.

Related: #4407

let validator_len: BalanceOf<T> = (validators.len() as u32).into();
let total_rewarded_stake = Self::slot_stake() * validator_len;

let (total_payout, max_payout) = inflation::compute_total_payout(
	&T::RewardCurve::get(),
	total_rewarded_stake.clone(),
	T::Currency::total_issuance(),
	// Duration of era; more than u64::MAX is rewarded as u64::MAX.
	era_duration.saturated_into::<u64>(),
);

[joepetrowski]

After talking with @rphmeier, we may just want to use the median amount at stake instead of the sum. Either way, we should not use the minimum as it may be highly variable era-to-era.

[rphmeier]

The research.web3.foundation site should describe this, if it doesn't already. Least-staked is OK, but bottom-tercile is probably the ideal metric

[gui1117]

This was fixed in lazy payouts PR