XLSX/XLS: pipe characters in cells break Markdown tables

A literal | in an Excel cell value or header is data, not a column separator, but it is emitted raw, so converted rows gain phantom columns and the Markdown table is corrupt.

Repro (markitdown[xlsx], current main):

import io
import pandas as pd
from markitdown import MarkItDown

buf = io.BytesIO()
with pd.ExcelWriter(buf, engine="openpyxl") as w:
    pd.DataFrame({"a": ["x|y"], "b": ["2"]}).to_excel(w, index=False)
print(MarkItDown().convert_stream(buf.getvalue(), file_extension=".xlsx").markdown)

Actual (2-column table, 3-column row):

## Sheet1
| a | b |
| --- | --- |
| x|y | 2 |

Expected: | x\|y | 2 |, the same escaping the CSV converter already applies (see #2019 / #2266). Same problem in the .xls converter.

Start with the provided pandas/openpyxl reproduction through MarkItDown().convert_stream, then compare the XLSX and XLS converters with the CSV converter’s escaping referenced in #2019 and #2266. Done means literal pipe characters in spreadsheet cells and headers are escaped in the generated Markdown table without adding phantom columns.