[Dev] Update from CTSolvers

[ocots]

In CTSolvers, some code must be in CTModels. Transfer it.

[ocots]

@JuliaRegistrator register

Release notes:

CTModels v0.7.0

Highlights

• New typed core for OCP models and solutions
  Split the old monolithic src/types.jl into a structured src/core/types/ hierarchy (ocp_components, ocp_model, ocp_solution, nlp, initial_guess, …). This clarifies the representation of models, solutions, constraints, and related metadata, and adds the alias AbstractOptimalControlProblem = CTModels.AbstractModel for better interop.

• New NLP modelling layer
  Introduced a dedicated src/nlp/ layer (problem_core.jl, discretized_ocp.jl, model_api.jl, options_schema.jl, nlp_backends.jl, …) to build NLP models from optimisation problems and OCP models, and to map NLP solutions back to CTModels.Solution. Adds support for ADNLPModels- and ExaModels-based backends as first-class CTModels components.

• Initial guess utilities
  New, typed initial-guess layer in src/init/initial_guess.jl: pre_initial_guess builds an OptimalControlPreInit container from raw user data (functions, vectors, scalars), and initial_guess builds and validates an OptimalControlInitialGuess against an AbstractOptimalControlProblem. Dedicated tests cover the new types and constructors.

• JSON / JLD I/O improvements
  Reworked JSON export/import in ext/CTModelsJSON.jl to handle infos::Dict{Symbol,Any} more robustly and predictably, with improved tests for JSON and JLD round-trips in test/io/test_export_import.jl.

• Plotting and examples
  Small improvements in the plot extension (ext/plot.jl, plot_default.jl, plot_utils.jl) and additional examples/tests for plotting and printing solutions in test/plot/test_plot.jl.

• Documentation overhaul
  New, more didactic index page in docs/src/index.md explaining CTModels' role in the OptimalControl/control-toolbox ecosystem, a new Interfaces section (docs/src/interfaces/), and reorganised API reference pages with improved automatic API doc generation using docs/docutils/DocumenterReference.jl.

• Extensive test suite
  Many new tests for core types, initial guesses, the NLP layer, I/O, OCP building blocks, and plotting (see the new subdirectories test/core/, test/nlp/, test/io/, test/ocp/, test/plot/, test/problems/). test/runtests.jl was refactored into a more modular structure.

---

Breaking Changes / Compatibility Notes

• CTBase compatibility bump
  compat for CTBase was raised from 0.16 to 0.17 in Project.toml. Downstream packages must be able to use CTBase ≥ 0.17 to upgrade to CTModels v0.7.0.

• New hard dependencies
  CTModels now declares additional dependencies in Project.toml:

  • ADNLPModels = "0.8"
  • ExaModels = "0.9"
  • NLPModels = "0.21"
  • SolverCore = "0.3"
  • KernelAbstractions = "0.9"
    Projects with tight compat bounds on these packages may need to update their compat entries.

• Internal file/layout refactor
  The internal layout of the source tree changed significantly:

  • src/types.jl was split into multiple files under src/core/types/,
  • src/init.jl was replaced by src/init/initial_guess.jl,
  • most OCP-related files were moved under src/ocp/,
  • NLP-related code was moved under src/nlp/.
    Publicly exported names have been kept as stable as possible, but code that relied on internal file structure or non-exported implementation details may break.

• JSON export/import behaviour
  JSON I/O has been tightened and made more structured. The intent is to be more robust, but very low-level consumers of the previous JSON format may see behavioural differences and should re-check their pipelines.

New Features

• Added a typed OCP core in src/core/types/, including ocp_components.jl, ocp_model.jl, ocp_solution.jl, and nlp.jl, to model optimal control problems and their solutions more explicitly.
• Added a new NLP layer in src/nlp/ with problem_core.jl, discretized_ocp.jl, model_api.jl, options_schema.jl, and nlp_backends.jl to interface CTModels with ADNLPModels and ExaModels backends.
• Introduced typed initial-guess types and constructors in src/core/types/initial_guess.jl and src/init/initial_guess.jl.
• Added JSON and JLD solution I/O helpers in ext/CTModelsJSON.jl and CTModelsJLD.jl to persist and reload solutions.

Enhancements

• Improved organisation of OCP-related code by moving files under src/ocp/ and sharpening the separation between core types, OCP model components, NLP layer, and I/O.
• Refined plotting support in the CTModelsPlots extension (plot.jl, plot_default.jl, plot_utils.jl) and aligned examples with the new types and solution structures.

Bug Fixes

• No additional user-facing bug fixes are explicitly highlighted in this release beyond those implied by the refactors and new tests. Please refer to the full changelog for low-level details.

Documentation

• Added a new, explanatory index page at docs/src/index.md describing CTModels' role and main concepts.
• Introduced an Interfaces section in docs/src/interfaces/ covering OCP tools, optimisation problems, optimisation modelers, and solution builders.
• Added a custom API reference generator in docs/docutils/DocumenterReference.jl and updated docs/make.jl to improve API documentation structure.

Tests

• Added extensive tests for core types in test/core/,
  including OCP components, models, solutions, NLP types, and utilities.
• Added tests for initial guesses in test/init/test_initial_guess.jl.
• Added tests for the NLP layer in test/nlp/, covering discretised OCPs, model API, backends, options schema, and problem core.
• Added I/O tests in test/io/ for export/import behaviour and extension errors.
• Added OCP-level tests in test/ocp/ and problem examples in test/problems/.

Internal

• Updated CI configuration in .github/workflows/ and added formatter.lock to track formatting state.
• Ignored profiling scripts via .gitignore and removed them from the tracked files.
• Reorganised the test suite to mirror the new core/, ocp/, nlp/, io/, meta/, and problems/ structure.

[JuliaRegistrator]

Registration pull request created: JuliaRegistries/General/143910

Tagging

After the above pull request is merged, it is recommended that a tag is created on this repository for the registered package version.

This will be done automatically if the Julia TagBot GitHub Action is installed, or can be done manually through the github interface, or via:

git tag -a v0.7.0 -m "<description of version>" e0129d19c2b624c288fa7694ed123658150b70a3
git push origin v0.7.0