Refactor and extend History class

[timmens]

• Remove HistoryArrays class and add these methods to History class
• Add timing module that contains functionality to create CostModels (as described Improve runtime measures for criterion plot and benchmarking plots #547)
• Add new methods and properties to History class, most importantly fun_data and params_data, which return tidy data frames of the function evaluations and a user-selected time measure.
• Check visually that plots have not changed!

Note

1. For the history processing for visualizations (history_plots.py), we must wait for Align plotting from logs and results #553
2. For the usage of the new methods fun_data and params_data in the visualization code we will also wait for Align plotting from logs and results #553

[codecov]

Codecov Report

Attention: Patch coverage is 97.39130% with 6 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/optimagic/optimization/history.py	96.51%	6 Missing ⚠️

Files with missing lines	Coverage Δ
src/optimagic/__init__.py	100.00% <100.00%> (+6.66%)	⬆️
src/optimagic/benchmarking/run_benchmark.py	92.30% <100.00%> (ø)
src/optimagic/optimization/convergence_report.py	100.00% <100.00%> (ø)
...agic/optimization/internal_optimization_problem.py	95.66% <100.00%> (+0.44%)	⬆️
src/optimagic/optimization/process_results.py	94.11% <100.00%> (-0.17%)	⬇️
src/optimagic/timing.py	100.00% <100.00%> (ø)
src/optimagic/visualization/history_plots.py	90.84% <100.00%> (+0.12%)	⬆️
src/optimagic/optimization/history.py	93.63% <96.51%> (+6.67%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[timmens]

The new plotly version (6.0.0) seems to have a bug. I had to restrict it. I could not find a MRE quickly. However, I already found out that the problem stems from these lines:

# plotting_utilities.py[100-106]
    if share_yrange_all:
        lb = []
        ub = []
        for f in plots:
            for d in f.data:
>               lb.append(np.min(d["y"]))
>               ub.append(np.max(d["y"]))

The problem seems to be that d["y"] (and for that matter also d["x"] a few lines below) return a dictionary instead of a list of floats.