docs: document four undocumented parameters in pipeline internals

[iridescentWen]

Follow-up to #12204 (merged), same defect class, different four sites.

What changed

Four functions each document every parameter except one. Added the missing :param: entry in
each case — docstrings only, no behavior change.

Location	Missing	Why it reads as an oversight, not a style choice
haystack/components/preprocessors/document_cleaner.py:262	text	The other three params (n_chars, n_first_pages_to_ignore, n_last_pages_to_ignore) are all documented, and the prose above already explains that pages in text must be form-feed separated — so the constraint was known, it just never became a param entry.
haystack/core/pipeline/base.py:1220	is_resume	component_name, component, inputs all documented. is_resume is load-bearing: at base.py:1241 it takes a different branch that returns the first available socket value instead of consuming inputs.
haystack/core/pipeline/draw.py:247	super_component_mapping	graph, server_url, params, timeout documented, plus both :raises: entries — only the last parameter is missing.
haystack/core/pipeline/pipeline.py:126	break_point	First five params documented plus :raises: and :return:. break_point is keyword-only, which is likely why it was missed when added.

How these were found

An ast pass comparing each function's signature against the names its docstring documents,
reporting only where the docstring is mostly complete (≥70% of params documented, ≤2 missing).
That threshold is the point: a function documenting nothing is a style choice, one or two
stragglers in an otherwise-complete list is an oversight.

Same method as #12204, which sjrl merged. Recall-checked before use: rolled the repo back to
5add74ec (the parent of #12204) and confirmed the scanner re-finds
LinkContentFetcher.__init__ missing ['request_headers'], one of the defects that PR fixed.

Scanning current main (5b791afa) reports exactly these four and nothing else; after this
change it reports zero.

Accuracy note

I verified each parameter's behavior in the function body rather than paraphrasing its name. One
draft was wrong and got corrected: I first wrote that break_point "saves a snapshot and raises",
but pipeline.py:150-154 only raises BreakpointException.from_triggered_breakpoint(...) before
the component runs — no snapshot is written there. The committed text says what the code does.

I also avoided putting a literal "\f" in the document_cleaner docstring: that docstring is not
raw, so the escape would embed an actual form-feed control character. The added line refers to the
form feed described in the prose above instead.

Validation

$ hatch run fmt
All checks passed!
549 files left unchanged

$ hatch run test:types
Success: no issues found in 408 source files

$ hatch run test:unit -k "document_cleaner or pipeline or draw"
540 passed, 5843 deselected, 3 warnings in 19.96s

No release note: the commit is prefixed docs:, which release_notes.yml exempts via its
git log --pretty=%s origin/main..HEAD | grep -E '^(ci:|docs:|test:)' check.

All four are private methods, so no public API surface changes.

---

🤖 Written with Claude Code. Every site was opened and read
individually; the commands above were run locally with the output shown.

[vercel]

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

A member of the Team first needs to authorize it.

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
haystack/core/pipeline
base.py
Project Total

This report was generated by python-coverage-comment-action

[anakin87]

Thank you!