docs/api-guide.md: "an actor occupies its whole worker" contradicts multi-actor workers #2152

docs/api-guide.md line 141 (v0.3.0) says: "requests are not consulted today (an actor occupies its whole worker)."

Since v0.2.0 (#1836), a worker hosts several actors at once: --max-actors defaults to 1000 (cmd/ateom-gvisor/main.go:83), and the scheduler places actors by remaining CPU and memory (cmd/ateapi/internal/scheduling/scheduling.go:153). docs/api-guide.md lines 66-67 and docs/glossary.md lines 48-50 already say so. On a v0.3.0 kind install with a two-replica gVisor WorkerPool, kubectl ate get actors showed two running actors on one worker pod at once.

The parenthetical at line 141 reads as the current placement rule and contradicts the other two passages. Suggest removing it, or restating it as how requests are treated now.