add persistency

[viktorbeck98]

Summary

• Introduces a new persistency module providing a flexible, extensible framework for storing and analyzing event data over time
• Implements multiple storage backends: pandas-based EventDataFrame for simple use cases and Polars-based ChunkedEventDataFrame with built-in retention and lazy materialization for streaming scenarios
• Adds a stability tracking system that classifies whether event variables are converging to stable values (STATIC, STABLE, UNSTABLE, RANDOM) using segment-based change analysis
• Includes supporting utilities: RLEList for memory-efficient run-length encoded sequences and preview_helpers for concise collection formatting

Architecture

The module follows a strategy pattern where EventPersistency orchestrates data ingestion and delegates to pluggable EventDataStructure implementations:

• Dataframes — raw event storage (pandas EventDataFrame, Polars ChunkedEventDataFrame with max-rows retention)
• Trackers — behavioral analysis via SingleTracker → MultiTracker → EventTracker hierarchy
• Stability tracking — StabilityClassifier divides a variable's change history into segments and applies increasingly strict thresholds to determine convergence

New dependencies

polars (used by ChunkedEventDataFrame), pandas (used by EventDataFrame)

Test plan

• test_persistency.py — 463 lines covering EventPersistency, EventDataFrame, ChunkedEventDataFrame, variable filtering, and edge cases
• test_stability_tracking.py — 554 lines covering RLEList, StabilityClassifier, SingleStabilityTracker, MultiStabilityTracker, and classification scenarios

Solves #15

[viktorbeck98]

@ipmach waiting for review

[ipmach]

Looks good!, I move the persistency tests to the test_utils folder because the persistency is inside the utils folder.

[ipmach]

@viktorbeck98 if you are ok with my last commit (moving test to the test_utils folder). Feel free to do the merge!