Save and quit on sigint and sigterm

[jlamypoirier]

✨ Description

Add a signal handler for training that catches interrupts. When a signal is received, the trainer finishes the current batch, then saves a checkpoint and stops gracefully.

This should be usually enough to save and quit within the allotted time, with some exceptions:

• The validation phase is not interrupted, so a batch with validation may not have time to save. Addressing this would require saving a validation state so we can resume mid-validation.
• Exports may take a while, so a batch with export may not have time to save.

Nevertheless, this PR is an improvement and should be good enough for #241.

🔍 Type of change

Select all that apply:

• 🐛 Bug fix (non-breaking change that addresses a specific issue)
• 🚀 New feature (non-breaking change that adds functionality)
• ⚠️ Breaking change (a change that could affect existing functionality)
• 📈 Performance improvement/optimization (improves speed, memory usage, or efficiency)
• 🛠️ Code refactor (non-functional changes that improve code readability, structure, etc.)
• 📦 Dependency bump (updates dependencies, including Dockerfile or package changes)
• 📝 Documentation change (updates documentation, including new content or typo fixes)
• 🔧 Infrastructure/Build change (affects build process, CI/CD, or dependencies)

[tscholak]

LGTM! thanks!