fix: iconVersionId wrong encode decode

[tomasciccola]

on src/lib/{encode,decode}-conversions.js when converting the iconVersionId for IconVariant we where encoding//decoding the string as 'hex'.
Since the strings has a slash ('/'), we where losing it when encoding the document.
Tests weren't failing since we weren't generating a valid versionId (without the /${index})

• Check if doing Buffer.from(string, 'utf-8') and buf.toString('utf-8') is safe

[achou11]

Not super sure about the changes here. Makes me wonder if the blobVersionId field should be derived similarly to how we derive versionId in the Common schema.

That has a links field, where a Link is { coreDiscoveryKey: Buffer, index: number }

[tomasciccola]

Makes me wonder if the blobVersionId field should be derived similarly to how we derive versionId in the Common schema.

Yeah, I think that makes a lot of sense. I made it those changes on the last commit but now I have a type error because Icon.blobVersionId is typed as optional on the protobuf typing.
I thought putting [(required) = true] makes the field required (in some sense...). Is there smth that I'm missing? 'cause giving that field a default value doesn't make much sense... @gmaclennan any pointers?