Update knot vector synthesis for approximate_curve and approximate_surface

[GarthSnyder]

Fixes #119 and #120. Includes tests to verify that knot vectors are well-formed, that every knot span includes at least one new data point, and that knot vectors generated for reversed curves are the inverses of those generated in the opposite direction.

[codecov]

Codecov Report

Merging #121 (d1dcaea) into 5.x-dev (8ae8b12) will increase coverage by 0.48%.
The diff coverage is 100.00%.

❗ Current head d1dcaea differs from pull request most recent head 607e1b5. Consider uploading reports for the commit 607e1b5 to get more accurate results

@@             Coverage Diff             @@
##           5.x-dev     #121      +/-   ##
===========================================
+ Coverage    47.18%   47.67%   +0.48%     
===========================================
  Files           36       36              
  Lines         6739     6739              
  Branches      1663     1663              
===========================================
+ Hits          3180     3213      +33     
+ Misses        3270     3236      -34     
- Partials       289      290       +1     

Impacted Files	Coverage Δ
geomdl/fitting.py	13.98% <100.00%> (+13.98%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[GarthSnyder]

@orbingol, have you reached an assessment of #120? Even if you want to address that issue in a different way, it might be helpful to merge the tests from this PR, which are fairly extensive. (Of course, they will fail without a fix for the underlying issue.)

[orbingol]

Looks like this was automatically closed when I deleted the 5.x-dev branch. Please base against 5.x and reopen as i cannot update base branch myself.

[GarthSnyder]

Oops, sorry about that. This is now PR #187.