v0.44.11

Improvements

• Previously, internal timing metrics grew continuously in memory the longer your instance ran, consuming over 500 MB on busy systems with many bridges and data flows. These metrics now use fixed-size histogram buckets that no longer grow with uptime, reducing steady-state container memory by roughly 30%