Switch Validator Bonded Tokens to using integers

[rigelrozanski]

Once the fee pool is implemented (which is the module which will mainly be uzing non-whole tokens) we will be able to switch the validator bonded tokens from decimal tokens back to whole tokens.

There are funny things which need to be considered if this switch is made (which it seems as though it should be made).

What to do with fractional worth of delegations shares

Once validators are using full atoms once again there is an interesting question of what to do with the fractional (sub)-atoms from during an unbonding.

The worth of delegator shares could all simultaneously be set to fractional worth after a slashing event.

Because the losses per withdraw of this is going to be very very small (aka less than a satoshi for an atom) it's not worth attempting to account for it per delegation (computational/code-complexity overhead not worth it). I'm thinking that any remainder from your withdraw should probably just stay with the validator (increasing the worth of other delegators tokens until it normalizes to the next integer)

seems like a reasonable solution, just a funny thing to have to pay attention to

[rigelrozanski]

NOTE Currently we are are burning the fractional atoms on unbonding (which I think is defensive and fine, also considering these are nano atoms)

---

So here is a possible constraint on this: We currently don't burn fractional atoms on redelegation, which means that validators power can be adjusted to being non-integer. If we did burn fractional atoms on redelegation, then I think we would be good on this situation.

[rigelrozanski]

Actually as a part of this issue - all places where fractional atoms are burned or moved should be outlined in the docs!