Test decoder output

[birktj]

These tests are based on the assumption that the current decoder is
correct and are to make sure that future changes to the decoder
wont break it.

This will fix #24

[HeroicKatora]

Since this has been open for a long time now I suppose it is a matter of confidence in the change? But there is already another accepted merge based on it. Reading the images in question with some known correct software (or convert with e.g. imagemagick) would also give groundtruths.

[Robzz]

I just verified the values with the following quick'n dirty python opencv script:

#!/usr/bin/env python3
import cv2
import numpy as np
from sys import argv

img = cv2.imread(argv[1], cv2.IMREAD_ANYCOLOR | cv2.IMREAD_ANYDEPTH)
try:
    rows, cols = img.shape
    print("Sum: ", np.sum(img))
except ValueError as e:
    rows, cols, channels = img.shape
    print("Sum: ", np.sum(img))

All 4 are OK, I suggest we merge 👍