Add merge key to pipeline

### Feature description

I would like to be able to use merge key without declaring a resource. Without it, using PK alone restricts the types of merges I can do

### Are you a dlt user?

None

### Use case

_No response_

### Proposed solution

user tried to
```python
info = pipeline.run(data2,
table_name="users",
write_disposition="merge",
merge_key='ID'
)
```

### Related issues

_No response_

Start by tracing the Python pipeline.run call shown in the example and how write_disposition="merge" is handled. Done means merge_key="ID" works without declaring a resource and enables the requested merge behavior; the relevant implementation path and tests still need to be identified.

Written by the indexing model from the issue text.