Enforce little endian when reading Numpy test data file

[olebole]

Numpy data (written with ndarray.tofile() are usually system dependent; at least they depend on the byte order. This makes the test_countrate_neg_leak() fail on big endian platforms (IBM zSystems/s390x). This patch enforces little endian on all platforms with np.fromfile() so the test passes there as well. With this patch, the package passes on s390x.

[codecov]

Codecov Report

Merging #336 (71430d2) into master (2a382d7) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #336   +/-   ##
=======================================
  Coverage   94.46%   94.46%           
=======================================
  Files          17       17           
  Lines        2096     2096           
=======================================
  Hits         1980     1980           
  Misses        116      116           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2a382d7...71430d2. Read the comment docs.

[pllim]

Thanks! LGTM but I'll let the CI run. Even though this look good, I might not come back to this in a few days.

Do you think it is beneficial to add a s390x job like astropy here or is that overkill?

[pllim]

p.s. Also let me know if you need a new release. Thanks!

[olebole]

I think that S390s is quite exotic, and the bugs that appear there might be taken as minor. Probably it is enough when they are fixed only after I found them by having a new Debian package; a Ci test is IMO not required. I applied this as a patch to the Debian package, which is fine for me. A new version is not needed.