Close dataframe column names are being treated as one by the LLM (#3611)
We are sending sample dataframe to LLM with df.head().
If the column names are close by, LLM treats two columns names as one,
returning incorrect results.

In the above case the LLM uses **Org Week** as the column name instead
of **Week** if asked about a specific week.
Returning head() as a markdown separates out the columns names and thus
using correct column name.
