fix #10 with the addition of a `transparency` option

[chrisdeely]

Uses a matchingPixelAlpha option for looksSame.createDiff to control the alpha of unchanged pixels.

This necessitated including the alpha channel in color comparisons in order to properly validate the diif images are rendered properly.

[kisenka]

@chrisdeely are you planning to update your PR according to requested changes? If not, I can do it :) This feature will be very handy!

[chrisdeely]

@kisenka I'll happily let you take the reigns here! I obviously have not gotten back to this in a timely manner

[drairi]

Could this provide a fix for https://github.com/gemini-testing/looks-same/pull/59 ?

[JoshuaM1995]

I know this is a few years old, but is there anyway to merge this in? I don't know much about the library but I'll help if possible. I'm working on a visual change detection app for websites and transparency would really help me!

[drairi]

@JoshuaM1995 that would be lovely! there has been some activity in this repo recently, so maybe it'll happen!

[DudaGod]

Hello guys. I will try to look at this on this week.

[chrisdeely]

i'm taking a look at this - the repo has moved on quite a bit since I initially made this PR, I'll see if I can adapt it to the new style

[chrisdeely]

I have updated my branch to re-implement this feature according to the new code style & features. the tests are all passing and the new matchingPixelAlpha option should allow users to visually remove matching pixels from the diff images.

Since the library now supports more image formats than just png, the alpha channel is defaulted to 255 if no value is present in the source image.

I have not tested the impact on antialiasing detection, but if there are issues there, please let me know and I'll try to help out.