feat: rewrite instruction and state types to support versioning

[bmuddha]

core record type now supports versioning via enum representation sync instruction also supports versioning to reflect changes in Record type evolution.

• some dependencies version adjustements

[GabrielePicco]

LGTM! The only consideration is that it may be worth adding a discriminator to the account, so that we can easily retrieve all the records by filtering for it. A follow-up consideration is that if we have a discriminator, the enum used for versioning could be avoided, as we could just bump the discriminator with the same effect.

[taco-paco]

Looks great! Just a small comment on .take usage in getters

[bmuddha]

LGTM! The only consideration is that it may be worth adding a discriminator to the account, so that we can easily retrieve all the records by filtering for it. A follow-up consideration is that if we have a discriminator, the enum used for versioning could be avoided, as we could just bump the discriminator with the same effect.

How borsh will be able to differentiate versions, if we don't use enums, every version bump will keep breaking existing code...
And borsh already encodes discriminant for the enum, so adding one explicitly is kind of redundant in this case
@GabrielePicco

[thlorenz]

Just minor stuff, but maybe you can address

[GabrielePicco]

How borsh will be able to differentiate versions, if we don't use enums, every version bump will keep breaking existing code...
And borsh already encodes discriminant for the enum, so adding one explicitly is kind of redundant in this case

They will just be treated as different accounts, since the discriminator is different, each parsing it's own struct. New version won't break previous ones.

What I'm thinking is that most likely we need to retrieve all records with something like:

let filters = vec![RpcFilterType::Memcmp(Memcmp::new(
    0,
    MemcmpEncodedBytes::Bytes(vec![1, 0, 0, 0, 0, 0, 0, 0]),
))];

But I see your point, if Borsh serialize the enum in a predictable way, in the first bytes of the accounts, we could use it as a discriminator. It's gonna be not standard (typically the discriminator is the first 8 bytes) and I guess variable size, but if the above is true, it's fine with me

[thlorenz]

Overall LGTM, but please make those error msgs a bit more helpful by including pubkeys :)