Reclassify pcov_pass (grid + samples) and add pcov_gate_class (samples), for every curve in a fitted result, using each curve's own LLOQ/ULOQ

Call this AFTER fit_calibration_freq()/fit_calibration_bayes() and AFTER compute_detection_limits_multiplate() – typically the very next line in the worker pipeline. Overwrites cr$grid$pcov_pass and cr$samples$pcov_pass for every curve, and adds cr$samples$ pcov_gate_class. calib_grid intentionally gets NO gate_class column (grid pcov_pass is recomputed for display symmetry only; it plays no role in deriving LLOQ/ULOQ, which happens upstream from the raw pcov profile before this function ever runs, so there is no circularity).

Usage

classify_pcov_gates_multiplate(mp)

Arguments

mp

A calibration_result or calibration_result_multiplate.

Value

The same class of object, with pcov_pass/pcov_gate_class (re)computed in place.

Details

Also computes each curve's inflection point exactly once and caches it at cr$detection_limits$inflection (list(x=, y=)), so flatten_and_save.R's diagnostics-row construction can read the SAME cached value instead of recomputing it independently – guarantees calib_diagnostics.inflect_x/y can never drift from whatever this function used for gating.