032 — Physically grounded window selection (production)

Production calibration-window selector. Same WindowSelector library class as 031 but with stricter physically-grounded thresholds (SNR floor, minimum cycle count per day, cross-chamber agreement requirement).

Output is the canonical 032_calibration_windows.csv consumed by the XPalm digital-twin calibration step (Julia notebook 040, out of scope for palmwtc 0.1).

Runs against the bundled synthetic sample but produces 0 windows (synthetic cycles are too short for high-confidence selection).

from palmwtc.config import DataPaths
from palmwtc.pipeline import run_step

paths = DataPaths.resolve()
print(paths.describe())

DataPaths (source=sample (bundled synthetic), site=libz):
  raw_dir       = /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/synthetic
  processed_dir = /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/Data/Integrated_QC_Data
  exports_dir   = /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/exports
  config_dir    = /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/config
  extras        = <none>

result = run_step("windows", paths)
print(f"Status:   {'OK' if result.ok else 'FAILED'}")
print(f"Elapsed:  {result.elapsed_seconds:.2f}s")
print(f"In:  {result.rows_in:>4} cycles")
print(f"Out: {result.rows_out:>4} windows")
print(f"Artefacts:")
for a in result.artefacts:
    print(f"  - {a}")

Status:   OK
Elapsed:  0.02s
In:     3 cycles
Out:    0 windows
Artefacts:
  - /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/exports/digital_twin/031_scored_cycles.csv
  - /home/runner/work/palmwtc/palmwtc/src/palmwtc/data/sample/exports/digital_twin/032_calibration_windows.csv

The two artefacts are the scored cycles (031-style output) and the identified windows (032 output). Real LIBZ data typically yields 50-200 high-confidence windows per month.