(re)Introduce scaling factor for mid

[Kolaru]

I am looking into updating IntervalRootFinding.jl for v0.22.

One think that disappeared that we need is scaled_mid.

This PR reintroduces it by giving mid an optional argument.

Technically it makes the default mid a bit slower (4 flops and 1 division, instead of 2 flops). I can split the default if it is somehow very critical for someone.

The new formula looks weird, but that's what we need to pass the test suite.

[codecov-commenter]

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (5909148) 83.50% compared to head (2dd229e) 84.07%.
Report is 3 commits behind head on master.

Files	Patch %	Lines
src/intervals/interval_operations/numeric.jl	88.23%	4 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #610      +/-   ##
==========================================
+ Coverage   83.50%   84.07%   +0.57%     
==========================================
  Files          24       24              
  Lines        2097     2091       -6     
==========================================
+ Hits         1751     1758       +7     
+ Misses        346      333      -13     

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

[OlivierHnt]

That's a good idea.

There probably were tests for scaled_mid that we can revive.

[OlivierHnt]

One note: we should add a check that α is between 0 and 1.

Perhaps something like

0 ≤ α ≤ 1 || return throw(DomainError(mid, "α must be between 0 and 1"))

[Kolaru]

I have added the domain check and some tests.