CI: run nightly workflow really nightly only

[kmilos]

The number of CI jobs on push/PR is simply unsustainable.

[ghost]

👇 Click on the image for a new way to code review

Legend

[codecov]

Codecov Report

Merging #2741 (cf8a24a) into main (4dfb781) will decrease coverage by 0.01%.
Report is 22 commits behind head on main.
The diff coverage is 55.55%.

❗ Current head cf8a24a differs from pull request most recent head 5144d96. Consider uploading reports for the commit 5144d96 to get more accurate results

@@            Coverage Diff             @@
##             main    #2741      +/-   ##
==========================================
- Coverage   63.91%   63.90%   -0.01%     
==========================================
  Files         103      103              
  Lines       22313    22313              
  Branches    10804    10805       +1     
==========================================
- Hits        14261    14260       -1     
- Misses       5828     5831       +3     
+ Partials     2224     2222       -2     

Files Changed	Coverage Δ
src/futils.cpp	73.09% <ø> (ø)
src/http.cpp	0.00% <0.00%> (ø)
src/makernote_int.cpp	65.91% <ø> (ø)
src/basicio.cpp	50.46% <36.36%> (-0.06%)	⬇️
include/exiv2/slice.hpp	90.56% <100.00%> (ø)
src/convert.cpp	52.35% <100.00%> (ø)

Pull request has been modified.

[neheb]

While we’re on the topic, these seem to be using make. Ninja is faster than GNU Make, specifically with CMake. I think I measured an 8% average speed up in my testing elsewhere.

[kmilos]

Ok, let's give it a go, I'll trigger it manually now to check.

[kmilos]

Sheeesh...

Ok, should be good to go now: https://github.com/Exiv2/exiv2/actions/runs/5799156458