Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

accumulatedMutationsInLineage resets on lineage switch (old behavior);
accumulatedMutations becomes a never-reset lifetime total.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…ors from genome nodes

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

The heavy evolution/lineage kernels were wall-clock throttled, so they ran
at run-to-run varying timesteps and perturbed the GPU execution timing of
subsequent simulation kernels, making runs non-reproducible and construction
tests flaky. They are now scheduled every 10 timesteps. Also restore
creatureIndex for losing threads in the dedup protocol.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…sEntry

Move EvolutionStatistics and the accumulated numCreatedCreatures/totalMutations
out of the bulk-copied StatisticsRawData into a standalone OverallStatisticsEntry
buffer on SimulationStatistics, analogous to the per-lineage entries. Also drop
the unused MutantStatistics diversity-tracking map, and make the Evolution
Dashboard timeline plots always render a filled area like StatisticsWindow.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Percent/min was extrapolated from whatever short window had accumulated
since sim start (or reload), producing huge values right after loading
a simulation. Now it waits for a full trailing minute of data and only
evaluates that minute; live history is also cleared on sim reset so a
reload can't be averaged together with the old run.