[Speedy]: fix ellipsis in ACAImage and squash warning about integer arrays

[javierggt]

Description

This PRs fixes issues found in python 3.11 (Speedy):

• ellipsis is not supported in slicing ACAImage
• a warning is issued in maude_decom when casting integer arrays (#166)

NOTE: while working on this, I found some unrelated issues I have not fixed. in this PR

Fixes #166 Fixes #165

Interface impacts

Testing

Unit tests

• [ ] No unit tests
• [x] Mac
• [ ] Linux
• [ ] Windows

Independent check of unit tests by @taldcroft

• [x] Mac

Functional tests

No functional testing.

[taldcroft]

@javierggt - my main question is if you understand why there are overflows (out of bounds) happening here that trigger the deprecation. Without understanding the code in detail, decom should by definition have predictable output values that are contained within well-specified bounds.

[taldcroft]

@javierggt - this needs a rebase now after #169. And hopefully the "fails independently of ..." comments will go away?

[taldcroft]

Unrelated to this PR, but for my future benefit can you add a comment line shown below.

From:

        out_rc = [None, None]  # New [row0, col0]

To:

        # These are new [row0, col0] values for the __getitem__ output. If either is left at None 
        # then the downstream code uses the original row0 or col0 value, respectively.
        out_rc = [None, None]

[javierggt]

my main question is if you understand why there are overflows (out of bounds) happening here that trigger the deprecation. Without understanding the code in detail, decom should by definition have predictable output values that are contained within well-specified bounds.

Yes.

The issue is not with the field size, but with converting from signed int8 to unsigned int8. I just took the easy way of replacing all instances where I was setting the dtype, although it was not necessary. Basically, I was converting numbers from np.int8 to np.uint8 by doing something equivalent to:

uval = np.array(val, dtype=np.uint8)

and changed it to this (based on their suggestion)

uval = np.array(val).astype(np.uint8)

One can reproduce this warning doing this:

from chandra_aca import maude_decom
import numpy as np
bits = b'\x8b'
bits = maude_decom._unpack("b", bits)
np.uint8(bits[0]).astype(np.int8)

Just now I looked at it again, and decided to fix in a different way so the relevant numbers do not require conversion, and I will fix it only where it is actually needed:

from chandra_aca import maude_decom
import numpy as np
bits = b'\x8b'
bits = maude_decom._unpack("B", bits)
np.uint8(bits[0]).astype(np.int8)

[javierggt]

@taldcroft I rebsaed on master, made some changes (mentioned in comments), and force-pushed.