Oracle Sqlalchemy destination varchar is created as clob

### dlt version

1.20.0

### Describe the problem

When I try to load data from Postgresql to Oracle using Sqlalchemy destination, all varchar fields are converted to CLOB type. When I try same operation with Sling Data, it correctly creates the varchar fields with lengths.

This is my code:
```
def load_psql_to_oracle() -> None:
oracledb.init_oracle_client()

source = sql_database(schema="public",
backend='pyarrow',
chunk_size=100000).with_resources("table1")
pipeline = dlt.pipeline(
pipeline_name="load_psql_to_oracle",
destination="sqlalchemy",
dataset_name="dest_schema",
progress="log"
)
load_info = pipeline.run(source,
write_disposition="replace"
)
print(load_info)
```

The table DDL for Postgrsql:
```sql
create table table1(id int, name varchar(10), surname varchar(20));
```

The table created by DLT on Oracle:
```sql
CREATE TABLE DEST_SCHEMA.TABLE1
(
ID NUMBER(19),
NAME CLOB,
SURNAME CLOB,
"_dlt_load_id" CLOB NOT NULL,
"_dlt_id" VARCHAR2(128 CHAR) NOT NULL
)
```

### Expected behavior

Column `name` should be `VARCHAR2(10)` and `surname` should be `VARCHAR2(20)`

### Steps to reproduce

1. Create a Postgreql table in Postgres: `create table table1(id int, name varchar(10), surname varchar(20));`
2. Insert dummy data: `insert into table1 values(5, "foo", "bar");`
3. Run the provided code above

### Operating system

Linux

### Runtime environment

Local

### Python version

3.10

### dlt data source

Postgresql (sql_database)

### dlt destination

_No response_

### Other deployment details

Destination: Oracle DB using sqlalchemy destination

### Additional information

_No response_

Start by reproducing the provided PostgreSQL-to-Oracle load with the SQLAlchemy destination, then trace the destination's column-type mapping for varchar fields. Confirm the fix by ensuring varchar(10) and varchar(20) produce VARCHAR2(10) and VARCHAR2(20) rather than CLOB.

Written by the indexing model from the issue text.