bug/RE_MULTISPACE_INCLUDING_NEWLINES was incorrectly used for Table or TableChunk

[JIAQIA]

🐞 Describe the bug

RE_MULTISPACE_INCLUDING_NEWLINES is applied to all elements of the Text category after partitioning PDF files. The relevant code is shown below:

out_elements = []
for el in elements:
    if isinstance(el, PageBreak) and not include_page_breaks:
        continue

    if isinstance(el, Image):
        out_elements.append(cast(Element, el))
    # NOTE(crag): this is probably always a Text object, but check for the sake of typing
    elif isinstance(el, Text):
        el.text = re.sub(
            RE_MULTISPACE_INCLUDING_NEWLINES,
            " ",
            el.text or "",
        ).strip()
        if el.text or isinstance(el, PageBreak):
            out_elements.append(cast(Element, el))

File path: unstructured/partition/pdf.py

However, if the element is a Table or TableChunk, the newline character "\n" is important and should not be removed in this context.

---

🔁 To Reproduce

1. Use partition_pdf on an image-based PDF that includes a table.
2. Observe that the newline characters within table content are removed by the above code.

---

✅ Expected behavior

Newlines should not be removed from Table or TableChunk elements.

---

🖼 Screenshots

If applicable, add screenshots to help illustrate the issue.

---

🧰 Environment Info

• OS version: macOS 13.6.7 (arm64)
• Python version: 3.10.14
• unstructured version: None
• unstructured-inference version: 0.7.36
• pytesseract version: 0.3.10
• Torch version: 2.3.0
• Detectron2: Not installed
• PaddleOCR: Not installed
• Libmagic version: libmagic 5.46 (bottled)
• LibreOffice version: 25.2.1

⚠️ Note: There were warnings about pip version checks failing.

---

📌 Additional context

The issue likely arises from applying the regex substitution to all Text elements indiscriminately, including those derived from tables, where "\n" conveys meaningful structure.