At the same time please support vector-valued dispersion values in DetectorMap (which will probably be needed anyway). The code can be as simple as this, but it needs to be a method. If we know more about the insides of the map we should be able to do better.

def getDispersion(detectorMap, fiberId, wavelength, dwavelength=0.1):
    """Return the dispersion in nm/pixel
    
    detectorMap: DetectorMap
    fiberId: `int`
       The fiber whose dispersion is desired
    wavelength: `float` or iterable or `float`
       Wavelengths where the dispersion is desired
    dwavelength: `float`
       Interval to use for finite differencing to determine the dispersion
    """
    wavelength = np.atleast_1d(wavelength)
    l0 = detectorMap.findPoint(fiberId, wavelength - 0.5*dwavelength)[:, 1]
    l1 = detectorMap.findPoint(fiberId, wavelength + 0.5*dwavelength)[:, 1]

    dispersion = dwavelength/(l1 - l0)

    return dispersion

pfsMerged now has units of summed counts per resampled pixel.

Before:

After:

A demonstration that the units of the pfsMerged is summed counts per resampled pixel:

At 637.4 nm:
Blue: 43411.88548364646 counts, dispersion 0.06598235465315838 nm/pixel
Red: 42618.45107528691 counts, dispersion 0.08778364248382786 nm/pixel
Merged: 93027.31531262455 counts, dispersion 0.08087158203125 nm/pixel
Note that (blue flux)/(blue dispersion) + (red flux)/(red dispersion) = (merged flux)/(merged dispersion)
(The error is < 1%. I’m guessing it’s not exact because there’s some variance weighting going on.)

Changes look fine to me.

Merged.