Invalid OID Handling in SQLAlchemy with asyncpg and autoload_with Causes DataError

Nobody has claimed this yet.

### Describe the bug

I encountered an issue when using SQLAlchemy with asyncpg and the autoload_with option while trying to load table metadata. The problem occurs because PostgreSQL's oid type is treated as a signed 32-bit integer (int32) by asyncpg, which leads to a DataError when the OID value exceeds the maximum range of int32 (2,147,483,647).

### Steps to Reproduce
1. Create a PostgreSQL table with an OID that exceeds the int32 range. For example, an OID like 3195477613.
2. Use SQLAlchemy to define the table with autoload_with to load metadata:

```python
from sqlalchemy import Table, MetaData

metadata = MetaData()
table = Table("master_product_view", metadata, autoload_with=conn)
```

3. Execute the code.

### Observed Behavior
The following exception is raised during metadata loading:

```python
sqlalchemy.exc.DBAPIError: (sqlalchemy.dialects.postgresql.asyncpg.Error) : invalid input for query argument $1: 3195477613 (value out of int32 range)
```

### Expected Behavior
The autoload_with option should correctly handle OID values, even if they exceed the int32 range.

### asyncpg/SQLAlchemy Version in Use

asyncpg 0.30.0
sqlalchemy 1.4.54

### Database Vendor and Major Version

PostgreSQL 14

### Python Version

3.13

### Operating system

Linux

No source file or test is named. Start by reproducing the SQLAlchemy Table(..., autoload_with=conn) metadata load against PostgreSQL 14 with asyncpg 0.30.0 and the reported OID, then trace the OID parameter handling. Done means metadata loading succeeds for OIDs above the int32 range without DataError.

Written by the indexing model from the issue text.