Each filter passes a different amount of light, which varies fairly smoothly with wavelength. L2 (near IR) images like this one are exposed in about a quarter-second. L7 (near UV) images are exposed for about 10 seconds.

Here's my analysis:

The speck doesn't appear in the next image; not a CCD flaw.

It doesn't appear in a later L2 image, so it's not a speck on a specific filter.

Digital artifacts tend to conform to the rectangular geometry of images. A spike ("point-source noise") will affect several pixels horizontally, depending on the duration of the spike relative to the scan/readout time of the sensor. Recurring spikes will give you a series of botched pixels down the image; if they're perfectly vertical they're probably somehow synced to the camera electronics themselves, othewise you tend to get diagonal herringbone patterns.

Non-perpendicular or non-periodic thingies like this one would probably have to have an outside cause. A cosmic ray might obliquely impact the sensor and take out several related pixels at an angle. Or the sensor might actually be seeing something...

Knowing the exposure time, it should be possible to calculate the proper motion of whatever-it-is based on its length. If you posit that it's at orbital altitude you have a distance from which you can get a velocity, and determine whether it's reasonable for an orbiting object. An ephemeris should reveal whether one of the moons would have been visible at the time.

Occam's Razor wants a mundane explanation, but there's little actual evidence to say whether it's mundane or extraordinary. Needs more work.