🐛 Make RL circuit sampling seed-aware

[flowerthrower]

🤖 AI text below 🤖

Description

Makes RL training-circuit sampling follow Gymnasium's standard seed lifecycle.

PredictorEnv now uses self.np_random, which is initialized and reseeded by Env.reset(seed=...), instead of a separate generator fixed to seed 10. Training-circuit paths are sorted before indexed sampling so equal seeds select equal circuits across filesystems.

Compiler-specific seeds, flat-policy defaults, ML estimator seeds, and experiment-only RNGs are unchanged because they already match the intended behavior or are outside this PR's scope.

This draft is stacked on #780. It intentionally contains only the raw seeding fix; documentation and new tests are deferred.

Validation

• repeated reset(seed=7) selects the same training circuit
• complete audit of seed and RNG sites under src/
• git diff --check

Checklist

• The pull request only contains focused changes required for this fix.
• New tests and documentation are intentionally deferred while the feature stack is assembled.
• I reviewed the complete base-to-head diff.

If PR contains AI-assisted content:

• Any agent that created, edited, or submitted GitHub content was explicitly authorized for that scope.
• Every agent-authored or agent-edited public text body begins with the visible disclosure 🤖 *AI text below* 🤖.
• I confirm that I have personally reviewed and understood all AI-generated content, and accept full responsibility for it.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/mqt/predictor/rl/helper.py	0.0%	2 Missing ⚠️
src/mqt/predictor/rl/predictorenv.py	0.0%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!