read_xlsx with "all_varchar = true" appends ".0" to integers.

I have an excel file with with strings ids (numbers, dots and hyphens, mostly).
Some of those ids are only numbers and, when reading the file, DuckDB appends a ".0" to those.

Is there a way to avoid this? I need the ids to stay unchanged.

For reference, I have a file with a columns like:

| reference |
| --------- |
| 12-3455 |
| 34-5678 |
| 12345 |

Doing `data = DuckDbQuery("SELECT * FROM read_xlsx(?, header = true, all_varchar = true)")` returns:

| reference |
| --------- |
| 12-3455 |
| 34-5678 |
| 12345.0 |

Start at the read_xlsx entry point and reproduce the issue with the sample reference values and all_varchar = true. Check how numeric-only Excel cells are converted to strings, then verify that values such as 12345 remain unchanged while hyphenated references still work as shown.

Written by the indexing model from the issue text.