feat: integrate & test FIP-0098 additions

[rvagg]

Ref: filecoin-project/go-state-types#376
Ref: filecoin-project/builtin-actors#1639

I've added some extras in our state types abstractions here, I'd like to add more and reduce (ideally remove) the direct uses of go-state-types versioned actor types we have through the codebase. (See #12960 for some of how annoying this is). My goal in the itest was to not refer to go-state-types at all while testing all of those additions. We should be able to approach almost all other uses of those types throughout the code in the same way, but we're pretty sloppy.

Also removed some batch balancer stuff.

[BigLep]

Requested review from @ZenGround0 given he gave approval on filecoin-project/builtin-actors#1639, but @ZenGround0 , if you can't look this week, please request @Stebalien instead.

[BigLep]

@rvagg : I assume this should be backported?

@rjan90 : would you ideally want to get this merged before the calibnet upgrade so migration pathways are flexed with this code? (I assume we'd just encourage a few calibnodes to use master with this change rather than cutting another RC at this point.)

[rjan90]

would you ideally want to get this merged before the calibnet upgrade so migration pathways are flexed with this code? (I assume we'd just encourage a few calibnodes to use master with this change rather than cutting another RC at this point.)

Yes, I think running either this branch / or master with the changes merged during the Calibnet upgrade will give us enough confidence to have this backported to the release/v1.32.0 branch

[Stebalien]

One suggestion but otherwise this looks correct.

[rvagg]

I also added a miner.PledgePenaltyForTermination utility wrapper for the v16 implementation while I was in there, but this one will error for everything prior to v16 since that's the simplified one we introduced.

[rvagg]

I shouldn't have merged this without a new go-state-types, I'll get that updated now in a new PR