fix peak picking corner case when an EIC has between 2 and 10 features

[hechth]

If an EIC after noise removal has between 2 and 10 features, it is always assumed to be a single peak no matter if it has 2 sections of consecutive scans at 2 distinct retention times. This causes very broad peaks instead of detecting 2 individual peaks.

recetox-aplcms/R/prof.to.features.R

Lines 936 to 960 in d6be955

	if (dplyr::between(num_features, 2, 10)) {
	# linear interpolation of missing intensities and calculate the area for a single EIC
	eic_area <- interpol.area(feature_group[, "rt"], feature_group[, "intensity"], base.curve[, "base.curve"], all_diff_mean_rts)
	rt_peak_shape <- c(median(feature_group[, "mz"]), median(feature_group[, "rt"]), sd(feature_group[, "rt"]), sd(feature_group[, "rt"]), eic_area)
	peak_parameters <- rbind(peak_parameters, rt_peak_shape)
	}
	if (num_features < 2) {
	time_weights <- all_diff_mean_rts[which(base.curve[, "base.curve"] %in% feature_group[2])]
	rt_peak_shape <- c(feature_group[1], feature_group[2], NA, NA, feature_group[3] * time_weights)
	peak_parameters <- rbind(peak_parameters, rt_peak_shape)
	}
	# application of selected model and method
	if (num_features > 10) {
	# find bandwidth for these particular range
	rt_range <- range(feature_group[, "rt"])
	bw <- min(max(bandwidth * (max(rt_range) - min(rt_range)), min_bandwidth), max_bandwidth)
	bw <- seq(bw, 2 * bw, length.out = 3)
	if (bw[1] > 1.5 * min_bandwidth) {
	bw <- c(max(min_bandwidth, bw[1] / 2), bw)
	}
	rt_profile <- compute_chromatographic_profile(feature_group, base.curve)
	if (shape_model == "Gaussian") {
	rt_peak_shape <- compute_gaussian_peak_shape(rt_profile, bw, component_eliminate, BIC_factor, aver_diff)
	} else {

[hechth]

The section handling this cases should simply be removed and the case for more than 10 features should simply become an else.