Simplify Migration Structs

[brennanjl]

This PR simplifies migration structs. There are three key changes:

1. It removes chain-id from the migration proposals, since they are effectively a suggestion anyways.
2. It simplifies the struct that is passed back from the client when querying for migrations. It previously used to return serialized genesis.json, but now returns a much slimmed down struct that only contains necessary info.
3. It adds kwild's minor version to the migration metadata when querying a node. This is because we still send the snapshot metadata serialized, and thus need to maintain compatibility between versions. I thought about moving snapshot metadata into core/types, but didn't to keep it simple.

[brennanjl]

Have not tested this btw

[jchappelow]

Wow this feels super clean. Have not tested either and we definitely want @charithabandi to review and test, but LGTM.

[brennanjl]

I'm going to test this over the weekend since it is such a big change

[charithabandi]

Looking into this rn

[jchappelow]

Regarding CI failure, on the "unhealthy" node1 I see:

2024-09-30 10:30:42 kwild version 0.9.0-pre (Go version go1.23.0)
2024-09-30 10:30:42 Root directory "/app/kwil"
2024-09-30 10:30:42 Private key path: /app/kwil/private_key
2024-09-30 10:30:42 2024-09-30T15:30:42.811Z    info    kwild   version 0.9.0-pre commit 
2024-09-30 10:30:42 2024-09-30T15:30:42.811Z    info    kwild   Entering migration mode {"migrate_from": "http://01d8aa26-5e5f-4756-95af-a6c614fb0aa2-node1-1:8484"}
2024-09-30 10:30:42 2024-09-30T15:30:42.818Z    info    kwild   Requesting genesis state from the old chain     {"listen_address": "http://01d8aa26-5e5f-4756-95af-a6c614fb0aa2-node1-1:8484"}
2024-09-30 10:30:42 2024-09-30T15:30:42.822Z    info    kwild   Genesis state not available     {"error": "status NoActiveMigration", "retry after(sec)": 30}

So it never gets healthy, at least not in 30 sec

[brennanjl]

Yeah we got it sorted, it was really really stupid.

I realized I was testing with a binary that was built from main, not this branch 🤦 . As soon as I tested with this branch I saw the error

[brennanjl]

Ok just finished testing this. I feel good about it

[jchappelow]

validateGenesisState could use an update. I don't think it should itself call failBuild (it's not in build.go and PrepareForMigration can return errors without having to panic). Also, a bunch of logic there is obsoleted now that we're ensuring it's existence here. The empty apphash logic in validateGenesisState is also a little questionable given the current changes, plus the bytes.Equal in the final return covers the length an nilness checks implicitly.

[brennanjl]

Oh yeah. Im just going to change this whole function since it is a bit convoluted in its usage

[jchappelow]

This doesn't need -- on activation-period and duration?

[brennanjl]

Oh whoops. Currently we do not test the cli for this, but yes you are correct

[brennanjl]

Is this good to go? @charithabandi @jchappelow

[jchappelow]

I am doing staging e2e this evening, but can merge. Any issues can be resolved after

[brennanjl]

Cool, thanks. Trying to get it merged b/c I know there will be merge conflicts with #1040

[jchappelow]

Cool, thanks. Trying to get it merged b/c I know there will be merge conflicts with #1040

They're minimal (I've resolved locally), but that's true