return BigNumber in the case of large number

[jcoveney-anchorzero]

I've confirmed that this fixes https://github.com/anchorzero/az/issues/2053

though I've also filed an issue on json-bigint to see how they will deal with it: sidorares#86 (this also explains the general issue, so is good context for this PR)

that said, I think the state of bigint is such that he will probably respond, but it's not that complicated of a function (the upside to all this is that I now understand what is going on much, much better!) and I think it's not the most actively maintained thing ever (eg he hasn't published it in ages)

one downside to this fix is that it exposes BigNumber, which is what they use. I think a next step would be to configure the output type, ideally to be AzDecimal or whatnot

[jcoveney-anchorzero]

also something relevant that I learned during this process..https://github.com/brianc/node-pg-types provides the ability to override the way in which specific types are parsed...but this won't fix the issue, because in this case, the parsing of a JSONB object is completely handed over to whatever the configured parser is--it doesn't use the node-pg-types parser. the upshot of this is that if we want to parse things into more custom types, we'd either need to make the JSONBigInt parser much fancier, or make our own parser. or, of course, use something that supports this...it looks like drizzle supports it, but I took a quick look at drizzle and there would be some work involved b/c it can't just introspect the database like zapatos does...though I saw someone had a project to generate the drizzle def based on your database, but it was just some random person's project.

that said, some things would be possible, especially if we abandon being able to round trip. for example, we could parse AzIds by extending the logic in json-bigint to "understand" the shape of these..but I don't know if that's a route we want to go. I'd rather either use a database interface like drizzle that understands the expected shapes, or find some other way to make parsing configurable on a per-table basis.

[jcoveney-anchorzero]

I don't think this is necesasry anymore