Look +/-1 for a match at each wavelength.

Use correlation: for target+small_noise, r is high; for large_noise
(or no match), r is lower

Try smoothing each curve and take derivatives until seeing a match:
filter dramatically with higher-order derivatives.

Look for patterns with same range, max-min, to identify similar peaks
and valleys.

Target shifted has same values, so look for matches of slices from
each spectrum, one shifted relative to the other.

Run ANOVA on all spectra to say whether all means are the same; if
reject H_0, run Tukey HSD to see which pairs are different: exclude
those! Repeat with larger alpha to exclude more.