Add voluntary exit via validator manager

[chong-he]

Issue Addressed

• Add --exit-epoch flag to validator-manager #4303
• Support voluntary exits via validator manager #4804

Proposed Changes

-Add voluntary exit feature to the validator manager
-Add delete all validators by using the keyword "all"

Additional Info

I reuse some code from the lighthouse account validator exit file:
https://github.com/sigp/lighthouse/blob/stable/account_manager/src/validator/exit.rs

Thank you @michaelsproul for the help and guidance along the way

[chong-he]

The --status is used to query the exit status of validators. The reason behind introducing it is:

In lighthouse account validator exit, the status was shown after a sleep of, e.g., 12s

lighthouse/account_manager/src/validator/exit.rs

Line 204 in 0d90135

sleep(Duration::from_secs(spec.seconds_per_slot)).await;

If we introduce a sleep here, then for each validator_to_exit, it will sleep for 12s, which is not really UX friendly. For practical consideration, the user's priority may just want to first publish the exit messages for all validators.

While there is a way to tweak it so that it only sleeps once and not for the rest of validator_to_exit (or shortening the sleep time), doing so will likely cause the exit status to be shown as:

Voluntary exit for Validator {} is waiting to be accepted into the beacon chain.

for the rest of validator_to_exit as the head block is not yet updated, which is not really useful either.

So I ended up introducing this --status flag to query the exit status separately. But I am not sure if this is a good idea.

Keen on hear the thoughts of reviewers on this.

[chong-he]

Not sure if this --merge is useful? I tested pasting the generated JSON on beaconcha.in and it doesn't seem to work. Maybe it is an issue with the generated JSON, or is it because submitting voluntary exit can only be done one at a time?