In-memory static figure export on the collect/plot path (PNG/SVG/PDF bytes)

[cursor]

Problem / context

collection.plot() is the right way for an app to get a Plotly figure, but there is no matching collect-level API to turn that figure into PNG / SVG / PDF bytes.

Collection.save(...) is data-only (parquet/csv/json/xlsx). The kaleido helper that exists — cellpy.utils.plotutils.save_image_files — lives outside collect, writes to disk, spawns a subprocess around fig.write_image, and prints status to stdout. That fits notebooks/scripts; it does not fit a desktop or FastAPI app that wants (bytes, media_type) for a download response (and a clear error when kaleido is missing).

Downstream (cellpy-simple-gui #27) calls fig.write_image directly today.

Spec

Something on the collect / plotting surface, e.g.

collection.to_image(\"svg\")           # -> bytes
# or
fig = collection.plot(...)
cellpy.plotting.write_image(fig, \"pdf\")  # -> bytes, raises if kaleido missing

Same formats users expect from kaleido (png / svg / pdf), in-memory, no subprocess or cwd side effects. Optional: Collection.save(..., formats=(\"svg\",)) that saves the plot next to the data — but bytes-first matters more for apps.

Acceptance criteria

• Public API returns image bytes (not only a filesystem path).
• Missing kaleido raises a clear, catchable error (no stdout-only status).
• Works for figures produced via collection.plot(...) / collected families.
• Docs/example show a minimal download-style usage.

---

Found while building cellpy-simple-gui on cellpy ≥2.1.1.post4. Full write-up: CELLPY_PAINPOINTS.md.