Add Control to Growth of the Staking Pallet

[shawntabrizi]

This PR attempts to address some of the issues related to scaling the Staking Pallet.

Due to constraints in the runtime, an unbounded and unlimited nominator and validator set is just not possible.

We introduce a few different layers of checks which will provide direct and indirect checks on the size of the staking system:

• We introduce a minimum active bond for both validators and nominators.

  • These minimum bonds are separate for the validator and nominator role, and are represented as storage items so they can be updated through governance.
  • When calling nominate or validate, we check that your current active bond meets these requirements.
  • When calling unbond we also check that your active bond does not drop below these requirements.
  • At any time, a user can chill, at which point they are able to unbond all their funds.
  • To keep things backwards compatible and safe as these bonds change, we introduce a chill_other extrinsic which is permissionless. This allows any third party user to point to an account where the minimum active bond is not satisfied, and chill that account.

• We introduce optional hard limits to the number of Validators and Nominators allowed in the staking system.

• tests for all the new logic

polkadot companion: paritytech/polkadot#3260

[kianenigma]

Looks overall good so far, but must have another look.

Idea: you can write a pre-migration function that does no migration, but only ensures that the current counter values are sane. Then we can put this in a runtime's custom migrations to have this auto-checked every now and then.

[kianenigma]

• I think we can provide a simpler and safe way for nominators to compete as well, based on their staked amount.
• The total limit of nominators should be a storage item that the governance can tweak more easily, without a full-fledged runtime upgrade. The validator count, meh, I care less for that.

Otherwise looks good to me.

[kianenigma]

@shawntabrizi please update this PR such that:

1. we put the counters in there
2. impose new MinStake and MinValidatorStake

but don't enforce any bounds yet. You can simply keep the functionality, but set it to None to indicate no bound is enforced now.

Then we can merge this PR asap as it would be simpler.

[shawntabrizi]

and this

[shawntabrizi]

This benchmark seems unused @kianenigma

[shawntabrizi]

Regarding me removing the lower bound on min-bond. I realized at this point in the PR (where things changed a bit in the logic) it is not needed, since it is impossible to have less than ED bonded, so this min is enforced anyway.

[kianenigma]

bot merge

[ghost]

Trying merge.