feat(schema_cache): column type + query

[juleswritescode]

Introduces an all-new, immediately-usable, delivered-straight-to-your-door data type in the SchemaCache: the Column.

The next step is to use that data type in the pg_completions crate – so we can suggest columns in a select clause, based on the queried table.

I wrote the query myself, with a little bit of help from ChatGPT and the postgres docs, so please make sure it follows best practices. 🙌🏻

Also wrote and renamed a test, and added full features for tokio on workspace level.

[juleswritescode]

Note that if we're using compound indexes, e.g. (user_id, tax_number), we'd have is_primary == true for tax_number.

This might be problematic for index suggestions etc. – wdyt?

[psteinroe]

I appreciate the effort to write a custom query, but what didn't you like from this one? :D

https://github.com/supabase/postgres-meta/blob/master/src/lib/sql/columns.sql

[juleswritescode]

@psteinroe Ah, I wasn't aware of the postgres-meta repo at all, I only took a look at this one... 🙈

We can use the postgres-meta query if you prefer, it's probably battle-proven.

But it does lack some info that seemed to me like we might need it in the future:

• the type_id which we could use to find the relevant type on the SchemaCache
• whether the column is a primary key
• the max length of the col if it's a varchar

Of course I'm not sure whether we'll need that data and whether the query fetches it 100% correctly. :)

Would you like me to use my query, the untouched postgres-meta query, or the postgres-meta query updated with those properties?

[psteinroe]

let's use yours and we keep it in mind when issues come up that it might be worth to go to postgrest-meta for it