Remove stub implementations for signed() and unsigned()

[kimikage]

This was separated from PR #350.

There was a discussion regarding signed(T) and unsigned(T) in late 2018, and as of at least 2020, no clear conclusion had been reached regarding generalization to types other than Integer (including the addition of signedtype() and unsignedtype()). (cf. JuliaLang/julia#30445)

More than five years have passed since then, but since there has been no movement toward generalization in other packages, I believe that even if this package supports SNormed or UFixed, FixedPoint would not support signed() or unsigned().

In other words, the stub has fulfilled its role as a placeholder. Furthermore, it will not affect existing codes, as long as they don't abuse the bug in Julia <v1.4.

---

I don't remember why I did it, but some of the sign tests should have been marked @test_broken instead of @test_throws.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.21%. Comparing base (6decae1) to head (b036653).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #356      +/-   ##
==========================================
- Coverage   97.22%   97.21%   -0.01%     
==========================================
  Files           7        7              
  Lines         792      790       -2     
==========================================
- Hits          770      768       -2     
  Misses         22       22              

☔ View full report in Codecov by Harness.
📢 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.

[kimikage]

Test results on julia v1.0 - v1.13 and nightly
https://github.com/JuliaMath/FixedPointNumbers.jl/actions/runs/38581189692

[PatrickHaecker]

I don't get the point of this PR. It breaks for older Julia versions exactly what #350 wants to avoid by being honest about what is supported and what isn't.

[PatrickHaecker]

On Julia 1.3, this breaks the code (no surprise here):

julia> using FixedPointNumbers; signed(1N0f8)
1

[kimikage]

Making code that was working stop working is a breaking change, but I’m skeptical about whether making code that wasn’t working start working in an undesirable way counts as a breaking change.

As far as these two methods are concerned, I don't have a strong opinion.
However, if we interpret “breaking changes” too strictly, we won't be able to fix bugs.

[PatrickHaecker]

Good point.

I still like the "more honest" approach of #350 better (and I would suggest that we avoid spending time on compatibility with versions no one uses with upgraded package versions anyway – sorry for triggering this), but I can follow your argument here: It wasn't working before and in that sense it's not too bad that it is not working now. Nevertheless, the question remains why it's better to return an incorrect result instead of throwing.

[kimikage]

I'd like to create a separate PR to fix the test, and then fix the bug of sign.