feat(go/adbc/driver/bigquery): Add extra table metadata

[serramatutu]

The driver wasn't returning a lot of the metadata available to the Go SDK.

This commit adds (most of) that data.

[lidavidm]

Given the discussion by other dbt engineers in #3449 maybe these values should be namespaced?

[serramatutu]

@lidavidm I was not aware of that discussion 😅 We opened this PR internally to unblock some use cases but happy to change the key names and upstream again if we wanna namespace or change the names in this.

[felipecrv]

@lidavidm I don't think everything should be namespaced, only keys that are likely to apply to many different drivers/systems.

[felipecrv]

@zeroshade are you happy with this?

[serramatutu]

I am not happy with this. But I'm not good enough at go to know a better way to ensure v is JSON-encodable and can be used with len(v). Happy to change if anyone has better ideas.

This at least compiles tho :P

[zeroshade]

Why not just:

encoded, _ := json.Marshal(v)
return string(encoded)

which should return either the encoded value or an empty string if it can't be encoded?

[serramatutu]

This doesn't preserve the current behavior:

if len(v) == 0 {
    encoded := ""
} else {
    encoded := json.Marshal(v)
}

If we just json.Marshal a value with length 0 we'll get [] instead of empty string.

We could change to simplify, but that would be a breaking change :D

[felipecrv]

@serramatutu can you skip setting the value on metadata altogether when its empty?

Going from

	metadata["Labels"] = encodeJson[map[string]string, string](md.Labels)

to something like this:

encodeJson[map[string]string, string](metadata, "Labels", md.Labels)

Helping reduce the bloat caused by keys in the metadata map and reducing the number of possible states to "either it exists with something or is not set at all".

[lidavidm]

I still think they should be namespaced, vs just dumping a bunch of values directly onto the schema (if the point is to prevent users from accidentally referring to the wrong metadata, which was the justification of namespacing some values in the first place, why doesn't that apply here?)