debug: correctly handle custom move blocking for LinModel

[franckgaga]

Closes #300, this one was hard ! 😮‍💨

This bug affected all controllers (LinMPC or NonLinMPC) and all transcription methods (SingleShooting and MultipleShooting). It was hitting when the model was a LinModel and using custom move blocking patterns (when Hc is a vector of integers). The resulting solution of the optimization problem was completely wrong.

The init_predmat method for SingleShooting and LinModel was not considering the fact that the decision vector $\mathbf{Z} = \mathbf{\Delta U}$ only include the free moves (thus it skips the blocked moves). The resulting E matrix is way more complex when I take this into account. I needed to revise the mathematical notation to keep track of all the various indices. The method init_defectmat for MultipleShooting and LinModel had a similar bug with the Eŝ matrix.

I added a test that compare the solution of custom move blocking patterns for LinMPC and NonLinMPC, both for SingleShooting and MultipleShooting.

The new mathematical notation is described here:

• init_predmat docstring (see extended help): https://juliacontrol.github.io/ModelPredictiveControl.jl/previews/PR301/internals/predictive_control/#ModelPredictiveControl.init_predmat
• init_defectmat docstring (see extended help): https://juliacontrol.github.io/ModelPredictiveControl.jl/previews/PR301/internals/predictive_control/#ModelPredictiveControl.init_defectmat

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.55%. Comparing base (81b2f31) to head (1a41588).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #301   +/-   ##
=======================================
  Coverage   98.55%   98.55%           
=======================================
  Files          28       28           
  Lines        4980     4993   +13     
=======================================
+ Hits         4908     4921   +13     
  Misses         72       72           

☔ 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.