pdfium outline is silently truncated at 15 levels

### Bug

`extract_outline_from_pdfium()` calls `pdoc.get_toc()` without arguments. pypdfium2
walks the outline recursively and defaults to `max_depth=15`, dropping every entry
below that level with only a log line:

```
Maximum recursion depth 15 reached (subtree skipped).
```

The other extractor in the same module, `extract_outline_from_docling_parse()`, walks
iteratively and is unbounded — `test_outline_deep_chain_does_not_raise_recursion_error`
asserts it handles 5000 levels. So the two extractors silently disagree on any document
whose table of contents is deeper than 15, and `PyPdfiumDocumentBackend` loses those
entries.

### Steps to reproduce

Build a one-page PDF whose outline is a chain 20 levels deep, then:

```python
import pypdfium2 as pdfium
from docling.utils.pdf_outline import extract_outline_from_pdfium

doc = pdfium.PdfDocument("deep_outline.pdf")
print(len(list(doc.get_toc()))) # 15 <- default bound
print(len(list(doc.get_toc(max_depth=128)))) # 20
print(len(extract_outline_from_pdfium(doc))) # 15 <- entries 16..20 are gone
```

Observed on both pypdfium2 4.30.0 and 5.13.0; the signature `get_toc(max_depth=15, ...)`
is the same in each.

### Docling version

2.124.0

### Python version

3.13.15

Start in docling/utils/pdf_outline.py with extract_outline_from_pdfium() and compare its pypdfium2 get_toc() call with extract_outline_from_docling_parse(). Run test_outline_deep_chain_does_not_raise_recursion_error and reproduce the 20-level outline case; done means the pdfium extractor retains entries beyond level 15 and the regression is covered.

Written by the indexing model from the issue text.