cellpy 2.0.0a5 — native-headers flip

Fifth cellpy 2 alpha. This release lands the native-headers flip: cellpy 2's runtime now speaks native cellpycore column names end to end — no legacy rename sandwich.

Highlights

• Default runtime is native (CellpyCell(native_schema=True)): from_raw/load → make_step_table → make_summary → save all run on native column names via the polars engine.
• Legacy compatibility preserved:
  • Legacy header attribute access (headers_normal.voltage_txt, …) is carried by a deprecation-warned shim.
  • native_schema=False keeps the full legacy bridge for v8 byte round-trips.
  • Legacy v8 files still load, and save(format="v8") still writes them, via to_legacy/to_native.
• Native merge() (campaign + continuation) works on the default path.
• Value-parity oracle: the native and legacy pipelines are asserted equal on 125 mapped columns (raw/steps/summary, incl. IR and CE) on the canonical dataset.

Breaking changes (alpha)

• data.raw / data.steps / data.summary now carry native column names by default (e.g. potential, cycle_num, datapoint_num, step_type). Access via the header objects (shim-carried) or migrate literals to native names. Set native_schema=False for the legacy names.
• get_cap curve frames use native CurveCols (potential/capacity/cycle_num/direction).
• cellpy.utils.easyplot removed (deprecated since 1.1).

Known gaps (tracked, xfailed)

• Native summary does not yet compute shifted_* capacity, RIC, or cumulated_coulombic_efficiency (#552).
• Native step_specifications application + exclude_step_types value parity to verify (#554).

Pinned to cellpycore==0.2.2.