Is your feature request related to a problem? Please describe.
The current CSVToDocument converter processes an entire CSV file as a single Haystack Document. This makes it difficult to handle use cases where each CSV row should become its own Document, which is common with structured datasets.

For example, if a CSV contains customer feedback with one entry per row, you may want to index each feedback item as a separate Document.

Describe the solution you’d like
Extend CSVToDocument to support row-level conversion, similar to DeepsetCSVRowsToDocumentConverter
(suggested by @sjrl).

Key points for row level conversion:

• Add a conversion_mode init parameter to choose whether the entire file or each row is converted into a Document (analogous to split_mode in CSVDocumentSplitter).
• For row-level conversion:
  • Allow selecting one column to populate Document.content (configurable by the user).
  • Store all remaining columns in Document.meta, using column names as keys and their values as the corresponding metadata values.

Describe alternatives you've considered
NA

Additional context
NA