Generate demo input files for the Constellab Bioprocess app.

This task takes a metadata-feature table (typically the output of CellCultureMergeFeatureMetadata) and generates the 4 input files required by ConstellabBioprocessLoadData:

1. Info CSV: One row per sample with Batch, Fermentor, Medium, and experimental conditions
2. Medium CSV: One row per unique medium with mean composition columns
3. Raw Data CSV: Synthetic time-series measurements (Biomasse, pH, temperature, NaOH) correlated with growth
4. Follow-up ZIP: Synthetic OD growth curves generated from logistic 4P model parameters (online monitoring)

Input Table Structure

The input table must have:

• Series column: sample identifier in format <Batch>_<Fermentor> (e.g., "M1_1")
• Medium column: medium name
• Medium composition columns (numeric, positioned before condition columns)
• Condition columns (e.g., "T ( C)", "pH (u_pH)", "OD at 880 nm", "NaOH (mL)")
• Model column (or param_* columns): marks the boundary between metadata and features
• param_y0, param_A, param_mu, param_lag: logistic 4P model parameters

Column Detection

• Metadata columns: all columns between Medium and Model (or first param_* column)
• Composition columns: metadata columns before the first one containing parentheses
• Condition columns: metadata columns from the first one containing parentheses onward

Generated Data

• Raw data: Biomasse (g/L) from logistic 4P model (offline measurements every 2h), pH decreases with growth, NaOH increases cumulatively, temperature stays constant (with noise)
• Follow-up: OD at 880 nm from logistic 4P model (online monitoring every 0.5h, with noise)