fix(color): improve precision in conversions and add tests

[vlakoff]

• Fixes a division by zero when converting rgb(0,0,0) to cmyk
• Improves precision in several places, by using:
  • corrected luminance coefficients
  • integer-based calculations to avoid floating-point precision loss
  • rounding instead of implicit truncation

This fixes #184.

Also added unit tests to cover rgb ↔ cmyk ↔ gray conversions.
Disclaimer: I hate writing tests, so these are fully AI-generated. Though, they work perfectly, and cover all the changes I've made in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.95%. Comparing base (c2fe9b3) to head (0f46d47).

Additional details and impacted files

@@             Coverage Diff              @@
##               main     #217      +/-   ##
============================================
+ Coverage     70.81%   71.95%   +1.14%     
- Complexity      994      996       +2     
============================================
  Files            49       49              
  Lines          3169     3170       +1     
============================================
+ Hits           2244     2281      +37     
+ Misses          925      889      -36     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[DASPRiD]

This looks generally good and I'm fine with AI generated tests, as long as they are user-reviewed to make sure they cover the cases as intended.

What I'm unhappy with are all the comments usually inserted by AI models. Code should usually be self-documenting, comments should only be required either in doc blocks to expand on the class/function names or inline if there are edge cases which cannot be expressed by the code itself.

Can you go over the diff and remove unnecessary comments?

[vlakoff]

I’ve reviewed these tests, and they’re solid — definitely better than if I had tediously written them myself by hand. Frankly, I could have skipped mentioning they were AI‑generated and nobody would have noticed.

As for the comments, I find them really useful to get an efficient overview of the code. For these test classes, I don’t even bother reading the implementation first, I just scan the comments — much more efficient.

That said, if you point out specific comments you think should go, I’ll consider removing them.