docs: prepare 2.1.2 release notes and re-execute the ICA and batch tutorials

[jepegit]

Release prep for v2.1.2: the changelog, plus the documentation errors found while reviewing it.

What changed

HISTORY.md. The flat [Unreleased] list is now a real ## [2.1.2] - 2026-08-09 section, grouped by theme, with a lead paragraph noting that the interim v2.1.1.post7 / v2.1.1.post8 tags and the v2.1.2a1–a4 pre-releases fold into it. Coverage was checked against git log v2.1.1.post6..master; nothing invented or dropped.

Two tutorials rewritten and re-executed. Incremental capacity analysis and Batch processing taught API removed in 2.1 and could not be run as written — the batch page failed on its very first configuration cell, since prms.Paths does not exist in 2.x.

The published .md pages are generated from the notebooks by dev/render_example_notebooks.py, so editing them directly would not have survived the next render. The fixes were made in the .ipynb sources; both notebooks were then executed, their plotly figures backfilled to PNG, and the markdown regenerated through the official script. Every code cell, table and figure on those pages is now real output.

Migrations applied:

Old	New
prms.Paths.<name>	cellpy.config.paths.<name>
collectors.Batch*Collector + .show()	cellpy.collect collectors + .plot()
ica.dqdv_cycle / dqdv_cycles / dqdv_np	ica.dqdv (three input forms)
y="dq"	y="dqdv"
get_cap().voltage	get_cap().potential
b.summaries.<col>	b.summaries.pivot(...) (long-format polars)

Executing the notebooks — rather than only reading them — surfaced three breakages that a code review had missed: paper01 has no ir_charge column (the second summary subplot now plots coulombic efficiency), get_cap returns potential not voltage, and b.summaries is a polars frame without attribute column access.

Other documentation fixes. The pandas-only framing in The fundamentals of cellpy (pandas per-cell, polars in cellpy.collect, parquet on disk); nine factual errors in the About loaders developers-guide section; and "deprecated, removal in 2.1" wording in the migration guides for API that has since been removed.

How to test

• uv run pytest -m essential — 703 passed, 1 skipped locally.
• Read the two rendered pages: Incremental capacity analysis, Batch processing.
• Re-render is reproducible and scoped: uv run --group docs python dev/render_example_notebooks.py leaves the other seven example pages byte-identical.
• The batch notebook writes no artifacts into the tracked docs/examples/batch_utility/out/ tree.

Notes for reviewers

• The *_1.png → *_0.png renames in the batch figures are output-index renumbering from re-execution, not lost figures — the page still has all eight.
• Notebook editors tend to drop the required metadata / execution_count keys from output records, which makes nbconvert refuse the file; the committed notebooks already had this defect and it was repaired.
• The separate, still-stale top-level examples/ notebook tree is not touched here — tracked in Top-level examples/ notebooks still use removed 1.x API #869.

Closes #866

Made with Cursor

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.