fix: write an empty XLSX cell as empty in the markdown table format

[L4XB]

Why:

XLSXToDocument renders the same empty cell two different ways depending on table_format:

XLSXToDocument(table_format="csv")       # ,A,B\n1,col_c,col_d\n2,True,
XLSXToDocument(table_format="markdown")  # |  2 | True  | nan   |

The CSV format writes an empty field. The markdown format writes the string nan, which reads as content — a retriever indexes it, and an LLM asked what is in that cell answers nan.

This is not hypothetical in this repo: the fixture sheet is literally called "Table Missing Value", and test_run_markdown pinned the nan:

assert (
    documents[1].content
    == "|    | A     | B     |\n|---:|:------|:------|\n|  1 | col_c | col_d |\n|  2 | True  | nan   |"
)

The documented escape hatch does not work either. table_format_kwargs is passed to to_markdown, and tabulate does have a missingval option, but it only applies to None: a NaN reaches tabulate's formatter as a number and is written out as nan, so table_format_kwargs={"missingval": "N/A"} changes nothing today.

What:

• Replace the empty cells with None before rendering (value.astype(object).where(value.notna(), None)), so tabulate sees a missing value rather than a float.
• Default missingval to "" in the resolved kwargs, so an empty cell reads as empty — the same as the CSV format. table_format_kwargs still wins, so {"missingval": "N/A"} now does what the docstring says.

Only the markdown branch is touched; the CSV branch, the hyperlink handling and the metadata are unchanged.

|    | A     | B     |          |    | A     | B     |
|---:|:------|:------|   -->    |---:|:------|:------|
|  1 | col_c | col_d |          |  1 | col_c | col_d |
|  2 | True  | nan   |          |  2 | True  |       |

Tests:

• test_run_markdown — the pinned expectation moves from nan to an empty cell. That assertion is the regression: on main the new expectation fails with

  - |  2 | True  | nan   |
  + |  2 | True  |       |
  

• test_run_markdown_missing_value (new) — table_format_kwargs={"missingval": "N/A"} reaches tabulate. It fails on main too, because the option had no effect there.

test/components/converters/test_xlsx_to_document.py
  main         2 failed, 15 passed
  this branch  17 passed

The other 15 pass unchanged on both sides, including every CSV-format test.

ruff check and ruff format --check are clean on both files. A release note is included.

[vercel]

@L4XB is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[github-actions]

Hi @L4XB, thanks for your interest in contributing to Haystack! 🙏

⚠️ You currently have 3 open pull requests in this repository (#12769, #12768 and this one). Our review capacity is limited, so please hold off opening more PRs until we've had a chance to review your first 2 open PRs. This helps us give each contribution the attention it deserves. Thank you!

This is an automated message to help us keep the review queue healthy.

[L4XB]

recheck