v0.3.34

[v0.3.34] - 2026-04-05

Added

• Added early public infrastructure for Modelica agent evaluation, OpenModelica integration, benchmark construction, repair-loop execution, and external-agent comparison scaffolding.
• Added tests and runnable utilities for the early experimental line where they remain part of the public codebase.

Changed

• Consolidated the internal v0.3.x experiment chain into this public phase closeout.
• Removed task-level metrics, lane-by-lane conclusions, artifact deep links, private benchmark routing, and detailed attribution traces from the public changelog.
• Kept the public record focused on capability areas rather than experimental play-by-play.

Validation

• Public validation is summarized at the phase level only.
• Detailed task counts, pass rates, branch/lane outcomes, and intermediate research conclusions are intentionally not published.