MIRI Shower Improvements

[stscijgbot-jp]

Issue JP-3562 was created on JIRA by Michael Regan:

The current MIR Shower flagging code as a couple of issues:

Reference pixels are treated the same as data pixels in the convolution used to detect showers. This can lead to false positive detections of showers.

The determination of the median value is not accurate for short integrations. The code should use all the integrations to determine the median difference of pixels.

Pixels that have DQ values of 5 (jump + DONOTUSE) or 6 (jump + saturation) are not masked before the convolution.