Pipeline changes for better handling of failures

[aditya0by0]

• Add a set-learn-false CLI subcommand to mark a learned/additional definition as learn_success=False in the learned-definitions pickle.
• Make compare_with_c3p.py robust to missing learned definitions and provide a default --c3p-json path.
• Adjust validator behavior to continue past per-class validation errors, and refine recursive learning behavior by having _learn() return a boolean.

[Copilot]

Pull request overview

This PR updates several inference/utility scripts to better handle validation failures, improve c3p comparison ergonomics, and add tooling for inspecting/adjusting learned-definition pickles.

Changes:

• Add a set-learn-false CLI subcommand to mark a learned/additional definition as learn_success=False in the learned-definitions pickle.
• Make compare_with_c3p.py robust to missing learned definitions and provide a default --c3p-json path.
• Adjust validator behavior to continue past per-class validation errors, and refine recursive learning behavior by having _learn() return a boolean.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
nl_2_fol/inference/utils/show_learned_content.py	Adds a new pickle-editing command (set-learn-false) for flipping learn_success flags.
nl_2_fol/inference/utils/compare_with_c3p.py	Avoids KeyErrors when learned definitions are missing and adds a default c3p JSON path.
nl_2_fol/inference/utils/calculate_c3p_stats.py	Introduces a new script to compute and print F1 distribution statistics from c3p_trust.json.
nl_2_fol/inference/learner/validator.py	Continues validation despite exceptions and summarizes the number of failures.
nl_2_fol/inference/learner/learn_definitions.py	Makes _learn() return a success boolean and uses it when handling missing predicates.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.