Outdated operation precedence.

Here's an expression: `SELECT ~1 + 2`. If you plug it into `PostgreSQL` interpreter, here's what the output looks like:
```
postgres=# select ~1 + 2;
?column?
----------
-4
(1 row)

postgres=# select ~(1 + 2);
?column?
----------
-4
(1 row)

postgres=# select (~1) + 2;
?column?
----------
0
(1 row)
```
As you can see here, and according to [operation precedence](https://www.postgresql.org/docs/current/sql-syntax-lexical.html#SQL-PRECEDENCE) listed in documentation, the `+` operation is more powerful than the `~` operation, and thus we have PGBitwiseNot(1 + 2). However, if we plug this into the parser, we get the following AST:
```
...
projection: [
UnnamedExpr(
BinaryOp {
left: UnaryOp {
op: PGBitwiseNot,
expr: Value(
Number(
"1",
false,
),
),
},
op: Plus,
right: Value(
Number(
"2",
false,
),
),
},
),
],
...
```
Which is implying `(~1) + 2` instead of `~(1 + 2)`. This should be fixed.

**UPDATE:**
In code, I found a reference to an old [PostgreSQL precedence table](https://www.postgresql.org/docs/7.0/operators.htm#AEN2026), using which the parser is built. However, this table is for a very old (`7.0`, released `May 8, 2000`) version of PostgreSQL. There latest version is `15.1`. Shouldn't this be updated?

Inizia leggendo la tabella di precedenza di PostgreSQL 7.0 a cui fa riferimento il parser e la documentazione corrente di PostgreSQL sulla precedenza collegata nella issue. Confronta l’AST del parser per `SELECT ~1 + 2` con il comportamento documentato e le tre query di esempio. Il lavoro è completato quando il parser applica la precedenza prevista e produce l’AST corrispondente.

Scritto dal modello di indicizzazione a partire dal testo della issue.