Bug in SIRS kernel subtraction

[stscijgbot-jp]

Issue JP-3915 was created on JIRA by David Law:

In testing why some new NIRISS sirskernel coefficients weren't improving the data quality as much as expected, I ran across a bug in the existing code.  The quadrant boundaries are not being computed properly in the apply_conv_kernel step; currently this says

data[:, chan * npix * 3 // 4:(chan + 1) * npix * 3 // 4] -= template[:, np.newaxis]

but this will overflow the array.  The 4th quadrant for instance is trying to correct pixel numbers 4608:6144 of a 2048 element array.

[stscijgbot-jp]

Comment by David Law on JIRA:

Fixed by #9267

[stscijgbot-jp]

Comment by Melanie Clarke on JIRA:

It looks like the current indexing does match what was originally provided in the notebook in JP-3622.  Timothy Brandt - can you comment?

[stscijgbot-jp]

Comment by Timothy Brandt on JIRA:

The indexing in the notebook from JP-3622 looks wrong; David Law's fix looks correct.  

[stscijgbot-jp]

Comment by Melanie Clarke on JIRA:

Fixed by #9267 (now merged)