Pyvrp

Prize-collecting!

* Add PyVRP/Instances submodule by N-Wouda in https://github.com/PyVRP/PyVRP/pull/216
* Route object by N-Wouda in https://github.com/PyVRP/PyVRP/pull/221
* Prize-collecting by N-Wouda in https://github.com/PyVRP/PyVRP/pull/213
* Prizes in neighbourhood calculation by N-Wouda in https://github.com/PyVRP/PyVRP/pull/223
* Add prizes to Individual str output by N-Wouda in https://github.com/PyVRP/PyVRP/pull/224

**Full Changelog**: https://github.com/PyVRP/PyVRP/compare/v0.2.1...v0.3.0

Fixes CD build.

**Full Changelog**: https://github.com/PyVRP/PyVRP/compare/v0.2.0...v0.2.1

TL;DR: Lots of quality of life changes, and further improvements to the codebase since the initial release of PyVRP a few months ago.

What's Changed
* Rename "Feasible objectives" to just "Objectives" in plot title by N-Wouda in https://github.com/PyVRP/PyVRP/pull/161
* Reduce coupling by N-Wouda in https://github.com/PyVRP/PyVRP/pull/162
* Initial submission configs by leonlan in https://github.com/PyVRP/PyVRP/pull/160
* Individual.make_random by N-Wouda in https://github.com/PyVRP/PyVRP/pull/165
* Remove args from clang-format in pre-commit by leonlan in https://github.com/PyVRP/PyVRP/pull/167
* GA should reject empty populations by N-Wouda in https://github.com/PyVRP/PyVRP/pull/172
* Reduce coupling in Population and GA now takes initial solutions by leonlan in https://github.com/PyVRP/PyVRP/pull/168
* Deterministic C++ implementation of SREX by leonlan in https://github.com/PyVRP/PyVRP/pull/175
* Remove PenaltyManager state from LS operators by N-Wouda in https://github.com/PyVRP/PyVRP/pull/177
* Remove vehicle capacity from penaltymanager by wouterkool in https://github.com/PyVRP/PyVRP/pull/186
* Implement CostEvaluator by wouterkool in https://github.com/PyVRP/PyVRP/pull/200
* Move route sorting inside SREX by wouterkool in https://github.com/PyVRP/PyVRP/pull/202
* Generalise binary tournament by N-Wouda in https://github.com/PyVRP/PyVRP/pull/183
* Separate distance and duration matrices by N-Wouda in https://github.com/PyVRP/PyVRP/pull/209
* BugFix: loadIndividual sets depot TWdata to zero by wouterkool in https://github.com/PyVRP/PyVRP/pull/211
* Initialize ProblemData with list of Clients by wouterkool in https://github.com/PyVRP/PyVRP/pull/206
* Sort granular neighbourhood by proximity by N-Wouda in https://github.com/PyVRP/PyVRP/pull/214
* Benchmarks for v0.2.0 by N-Wouda in https://github.com/PyVRP/PyVRP/pull/220


**Full Changelog**: https://github.com/PyVRP/PyVRP/compare/v0.1.0...v0.2.0

First public release

PyVRP is a Python package that implements Hybrid Genetic Search as a state-of-the-art Vehicle Routing Problem (VRP) solver. The package is designed for the VRP with Time Windows (VRPTW), but can be easily extended to support other VRP variants. PyVRP combines the flexibility of Python with the performance of C++, by implementing performance critical parts of the algorithm in C++, while being fully customisable at the Python level. PyVRP is a polished implementation of the algorithm that ranked 1st in the 2021 DIMACS VRPTW Challenge and, after improvements, ranked 1st on the static variant of the EURO Meets NeurIPS 2022 Vehicle Routing Competition. The code follows good software engineering practices, and is well-documented and unit tested. PyVRP is freely available under the liberal MIT license.

Use M_PI because CIBW uses GCC9 for musllinux wheels, and `std::numbers::pi` is not yet supported by that compiler.

Pre-release version that now actually compiles on Windows