fetch_dataframe returning float types incorrectly

The **fetch_dataframe** method provided by the Redshift connector appears to return more decimal places than expected. For example, 0.074 becomes 0.07400000095367, likely due to how Redshift stores and retrieves floating-point or decimal values with higher precision than originally intended.

The first image shows what was returned by the **fetch_dataframe** method and although the revenue growth column is exactly what we expected (exact user inputs without more decimal places), when we actually take a look at one of the values specifically (second one in this case), we get the value in the second image. This suggests that fetch_dataframe is already returning extra decimal places but just not viewable when printing the whole dataframe. For now, we have identified that this is a problem for the datatypes **REAL** and **DOUBLE PRECISION** on the redshift connector.

Start by tracing the fetch_dataframe method and reproducing the reported values for REAL and DOUBLE PRECISION in Redshift. Compare the returned values with the input precision and existing connector behavior; done means the expected decimal values are returned without unintended floating-point digits, with a regression test covering both types.

由索引模型根据 Issue 内容生成。