Deprecate fixed but unknown

[mhovd]

As we have discussed, NPAG is not suitable for fixed but unknown parameter values.
As such, this functionality is removed, and as a result the API becomes somewhat cleaner.

This closes #116

[Copilot]

Pull Request Overview

This pull request removes the support for fixed-but-unknown parameter values (no longer requiring a fixed flag) in order to simplify the API. Key changes include:

• Eliminating the fixed boolean flag from Parameter and related methods.
• Updating all functions, generators, and tests to use the new Parameters API without fixed parameters.
• Removing unused ranges parameters from methods such as suggest_point and check_point.

Reviewed Changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/onecomp.rs	Updated test to remove fixed parameter input and set progress flag.
src/structs/theta.rs	Replaced separate random/fixed parameters with a unified parameters field.
src/routines/settings.rs	Removed fixed flag from Parameter and updated Parameters::add accordingly.
src/routines/initialization/sobol.rs	Removed filtering based on fixed flag, now using all parameter ranges.
src/routines/initialization/mod.rs	Simplified Theta construction by using parameters clone.
src/routines/initialization/latin.rs	Updated Latin hypercube generation to use the new Parameters API.
src/routines/expansion/adaptative_grid.rs	Adjusted method call to check_point by removing the ranges argument.
src/algorithms/npod.rs	Removed ranges field and updated suggest_point calls accordingly.
examples/* and benches/*	Updated examples and benchmarks to reflect the new Parameters API.

Comments suppressed due to low confidence (2)

src/routines/expansion/adaptative_grid.rs:47

• Ensure that the updated check_point method in Theta correctly retrieves and applies parameter ranges via parameters.ranges(), so that its behavior matches what was previously provided by the explicit ranges argument.

.filter(|point| theta.check_point(point, min_dist))

src/algorithms/npod.rs:361

• Confirm that removing the ranges argument from suggest_point does not adversely affect NPOD's optimization process, and that Theta::check_point uses the correct parameter ranges from the unified Parameters object.

self.theta.suggest_point(cp.to_vec().as_slice(), THETA_D);

[github-actions]

Bencher Report

Branch	deprecate-fixed
Testbed	orion

Click to view all benchmark results

Benchmark	Latency	Benchmark Result seconds (s) (Result Δ%)	Upper Boundary seconds (s) (Limit %)
bimodal_ke_fit	📈 view plot 🚷 view threshold	18.29 s (-2.47%) Baseline: 18.75 s	19.98 s (91.54%)

🐰 View full continuous benchmarking report in Bencher