Implement filtered autocomplete for algorithm selection

[janosg]

Short description

After this PR, algorithms can for example be selected as follows:

import optimagic as om
import numpy as np 

om.minimize(
    fun=lambda x: x @ x,
    params=np.arange(3),
    algorithm=om.algos.Bounded.scipy_lbfgsb,
)

After typing om.algos.Bounded, the autocomplete in any modern editor will only suggest algorithms that support bound constraints.

Categories that can be used to filter algorithms are:

• GradientBased
• GradientFree
• Global
• Local
• Bounded
• LinearConstrained
• NonlinearConstrained
• Scalar
• LeastSquares
• Likelihood
• Parallel

Categories can be selected in any order. We make sure that mutually exclusive categories (e.g. Local and Global) cannot be selected at the same time.

Detailed description

The changes are described in EP-02: Static Typing

Deviations from the EP

• The Unbounded and Unconstrained categories were removed because people rarely want to specifically choose an optimizer that does not support constraints or bounds.
• om.algorithms was renamed to om.algos because it is still unambiguous and makes the code shorter, which hopefully improves adoption of using this instead of strings for algorithm selection.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Files with missing lines	Coverage Δ
src/optimagic/__init__.py	100.00% <100.00%> (+6.89%)	:arrow_up:
src/optimagic/algorithms.py	85.58% <ø> (-14.42%)	:arrow_down:
src/optimagic/config.py	70.96% <100.00%> (+0.47%)	:arrow_up:

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[janosg]

Tests are failing due to the release of DFO-LS 1.5.1; This is unrelated to the current changes and would affect all currently open PRs (@timmens, @mpetrosian).