Plotting and Fitting: Handling of Noise

[cvallee]

Ideally we should check if "baseline" is noisy or not for each channel (Blank, Sample, ControlSample, and DMSO Cal.). We could simply analyse the standard deviation of the baseline (generally the measurement from 0 to load sample step), if it's higher than a specific threshold => too noisy. Threshold would have to be calibrated based on different sensograms.

Eda would like the following behaviour regarding how we should handle noise:

1. Blank noise: if Blank is too noisy, ignore the current Blank data and use data of the previous Blank for the Blank response subtraction step.
2. Sample noise: if Sample is too noisy, add a secondary heuristic "Noisy". Eda will repeat experiments for the samples labelled "Binding:Noisy" (or by checking the .csv - add a "Noise" column with values like Yes/No, True/False, Noisy/NaN).
3. ControlSample noise: if ControlSample is too noisy, ignore.
4. DMSO Cal. noise: if CycleType == "Binding" check DMSO noise. If too noisy, ignore the current DMSO Cal. data and use data of the previous DMSO Cal. for the fitting.