Cache interpretation of trees and avoid excessive recomputation of predicates/trees

[Jekannadar]

Adds a cache to the Tree class, _interpreted. Trees are immutable, as long as the interpretation is deterministic, having a cache on the Tree itself is fine.

Modify _generate_new_trees:

Currently, whenever it is called, it combines all arguments with all parameters, and then later duplicates are thrown away downstream.

_RuleProgress introduces some simple tracking, making sure to only compute new combinations:

• new params with new args
• new params with old args
• new args with old params

This gets some speedup across all CoSy tests and benchmarks, CoSy-Luigi speeds up a bit, and on projects that are basically using CLS to compute a powerset, this gets a very heavy speedup.

Note:

• For non-deterministic interpretations and predicates, this definitely is breaking. However, I could not come up with a use-case for those, you wouldn't use combinatory logic if you wanted to go to the Casino during inhabitation.
• The dict containing the _RuleProgress objects is keyed on the id of RHSRule instead of itself. While RHSRule is hashable, its hash is rather expensive, and Pythons built-in id works just the same for this use-case.

[tudo-seal-workflows]

Benchmark CoSy

Details

Benchmark suite	Current: b5a8825	Previous: be71b5d	Ratio
benchmarks/test_benchmark_maximal_elements.py::test_benchmark_maximal_elements	10.43306182813801 iter/sec (stddev: 0.00022465880145587798)	9.608281351562356 iter/sec (stddev: 0.010017085420616013)	0.92
benchmarks/test_benchmark_maze.py::test_benchmark_maze	4.359626682760569 iter/sec (stddev: 0.020393723606656953)	3.9235166334572096 iter/sec (stddev: 0.018797551923491352)	0.90
benchmarks/test_benchmark_maze_contains.py::test_benchmark_maze_contains	4.110375478673194 iter/sec (stddev: 0.023216042955672603)	3.5391214334611916 iter/sec (stddev: 0.025877608823890046)	0.86
benchmarks/test_benchmark_maze_loopfree.py::test_benchmark_maze_loopfree	4.1489713872557 iter/sec (stddev: 0.024998383997361814)	3.8477221717914887 iter/sec (stddev: 0.02072136799063595)	0.93

This comment was automatically generated by workflow using github-action-benchmark.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 95.89041% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 73.02%. Comparing base (27f28d7) to head (b5a8825).
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
src/cosy/core/solution_space.py	88.88%	3 Missing and 3 partials ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop      #91      +/-   ##
===========================================
+ Coverage    71.99%   73.02%   +1.03%     
===========================================
  Files           37       39       +2     
  Lines         2810     2944     +134     
  Branches       476      494      +18     
===========================================
+ Hits          2023     2150     +127     
- Misses         683      686       +3     
- Partials       104      108       +4     

Flag	Coverage Δ
macos-latest-3.10	72.92% <95.89%> (+1.04%)	⬆️
macos-latest-3.11	72.99% <95.89%> (+1.14%)	⬆️
macos-latest-3.12	72.99% <95.89%> (+1.14%)	⬆️
macos-latest-3.13	72.79% <95.89%> (+0.83%)	⬆️
ubuntu-latest-3.10	72.82% <95.89%> (+1.26%)	⬆️
ubuntu-latest-3.11	72.99% <95.89%> (+1.10%)	⬆️
ubuntu-latest-3.12	72.99% <95.89%> (+1.03%)	⬆️
ubuntu-latest-3.13	72.99% <95.89%> (+1.14%)	⬆️
windows-latest-3.10	72.92% <95.89%> (+1.25%)	⬆️
windows-latest-3.11	72.99% <95.89%> (+1.03%)	⬆️
windows-latest-3.12	72.99% <95.89%> (+1.03%)	⬆️
windows-latest-3.13	72.99% <95.89%> (+1.14%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.