Why have we decided to use lossy here and lossless in the previous diff?

According to the Android Developer link @atul shared:

Lossless encoding is only available if your minSdkVersion is set to 18 or higher.

I tried opening the "Converting Images to WebP" dialog by right-clicking on splash_background.jpg, but it says "Current minSdkVersion is unknown:"

The project's minSdkVersion is defined in the top-level build file:

Looks like it's 21, so not sure why it's unknown. If we can figure out how to set the minSdkVersion correctly so that the "Converting Images to WebP" dialog no longer thinks it's unknown we should be able to do lossless compression. Otherwise, I think lossy is the only way to go (although I don't know if the quality is that different, but as you said it's just a bunch of dots so it probably didn't suffer too much at 95%).

Just because I was curious, I looked up some JPEG vs WebP statistics and found this study. The conclusion:

We observed that the average WebP file size is 25%-34% smaller compared to JPEG file size at equivalent SSIM index. The SSIM vs bpp plots showed that WebP consistently required less bits per pixel than JPEG for the same SSIM index. These results indicate that WebP can provide significant compression improvements over JPEG.

So quality-wise we should be good. And I agree with your idea of generating the dots on the client-side, that would save space.

Edit: actually trying to go through with the lossless conversion for a JPEG file on Android Studio results in this error:
{F85902}

But lossless conversion works for PNG to WebP in Android Studio, even though the minSdkVersion is unknown. Not sure why, but that's probably why @atul did lossy conversion in this diff and lossless in the last one.

Is the difference in quality noticeable in any way? Why have we decided to use lossy here and lossless in the previous diff?

Any possible difference was imperceptible to me even at high zoom levels (thought I'm guessing it's the color gradient vs the "sharpness" of the "stars" where most of the compression is happening).

Why have we decided to use lossy here and lossless in the previous diff?

Lossless PNG => WEBP conversion still brought down file sizes a good amount and the files were pretty small to begin with.

"Lossless" JPEG => WEBP conversion didn't affect file sizes, whereas even with a quality of 95 we were able to reduce file sizes by ~90%.

If we want to save even more space, we can probably store only the positions and sizes of dots and generate the images on the client.

We probably could do something like that, but I'm not sure how we would go about doing that. I know we do some more involved stuff with the splash screen already (iOS vs Android, dev vs release, something about expo), so I'd guess it's a lot more effort than it's worth for marginal space saving.