Combine tern and pg-schema-diff

Over time, multiple migrations makes life hard to get a clear view of the actual database schema. I was wondering whether the schema could keept as the "desired schema" and use `pg-schema-diff` to generate the diffs meant to be applied by tern.

For example, at the very beggining, the `schema.sql` looks like this:
```sql
create table t1(
id serial primary key
);
```

Given that the schema is empty, the auto-generated migration should look like this:
```sql
create table t1(
id serial primary key
);
---- create above / drop below ----
drop table t1;
```

Then, someone adds a column to `t1` byediting `schema.sql` file:
```sql
create table t1(
id serial primary key,
placebo varchar not null
);
```

The auto-generated migration should look like this:
```sql
alter table t1 add column placebo varchar not null;
---- create above / drop below ----
alter table t1 drop column placebo;
```

This way, `schema.sql` reprensets the desired state and each migration is automatically generated using `pg-schema-diff`. Notice that both "create above" and "drop below" can be generated using `pg-schema-diff` by inverting from/to flags.

To make this very robust, an `pglite` or similar instance could be used.

Any thoughts?

Start with the proposed schema.sql as the desired schema, then review how tern currently generates migrations and how pg-schema-diff could supply the forward and reverse SQL. Consider whether a pglite or similar instance is needed for comparison. Done means schema changes produce reversible migrations in tern's expected format.

Written by the indexing model from the issue text.