Eth2.0-specs: Validator status transition issues

[hwwhww]

Issues

• [x] 1. The current spec doesn't describe how PENDING_EXIT transits to EXITED_WITHOUT_PENALTY. A draft diagram:
  
  (Please let me know if anything incorrect or unclear.)

• [x] 2. Potential min_empty_validator_index() issue: follow up this comment. Agreed with @djrtwo. IMO It doesn't make sense to allow the validators who have sent voluntary exit/withdraw message to make a top-up deposit.
  edited: just noticed the second case: involuntary ACTIVE -> PENDING_EXIT when v.balance < MIN_BALANCE.

• [ ] 3. COLLECTIVE_PENALTY_CALCULATION_PERIOD and "withdrawal period": In the codes, we use COLLECTIVE_PENALTY_CALCULATION_PERIOD while it is described as "withdrawal period" everywhere.

• [x] 4. get_active_validator_indices returns both ACTIVE and PENDING_EXIT validators.

  def get_active_validator_indices(validators: [ValidatorRecord]) -> List[int]:
      """
      Gets indices of active validators from ``validators``.
      """
      return [i for i, v in enumerate(validators) if v.status in [ACTIVE, PENDING_EXIT]]
  

  Maybe we should rename ACTIVE status to another name?

Proposed solution

For issue 1

1. @djrtwo suggested: https://github.com/ethereum/eth2.0-specs/pull/185/files#r237660001
   get_changed_validators() needs to check if state == PENDING_EXIT and then immediately set state = EXITED_WITHOUT_PENALTY.
2. Add the stub of the actual withdraw. For testing withdraw, maybe we can actually set v.balance=0 and note that the balance will be transferred to shard chain in phase 1.

---

potentially related to #198?

[djrtwo]

1, 2, and 4 addressed in #228

[djrtwo]

Closing. The rest of the penalty issues are reference in a specific penalty issue #196