[1.18] Fix actor drain timeout and placement dissemination docs

[JoshVanL]

The actor runtime configuration page claimed drainOngoingCallTimeout defaults to 60 seconds; the runtime default has been 2 seconds since v1.17.0 (DefaultOngoingCallTimeout in pkg/actors/api/config.go). The drain timeout clamp also shipped in v1.17.7, not v1.18 as stated.

Rewrite the drain timeout clamping section to distinguish the two dissemination timeouts: the daprd-side --actors-disseminate-timeout (default 30s, which the clamp compares against) and the Placement service --disseminate-timeout (Helm dapr_placement.disseminateTimeout, default 8s, the deadline after which Placement kicks a non-responsive sidecar from the round). Add guidance to keep drain values below the Placement dissemination timeout, since values that pass the clamp can still delay the table update acknowledgement past the 8s deadline.

Correct the kubernetes-production description of disseminateTimeout: it is the dissemination round failure deadline, not a dissemination delay; the default is 8s, and there is no enforced 1s-3s range.

Document the --actors-disseminate-timeout daprd argument and the dapr.io/actors-disseminate-timeout annotation, which were previously missing from the arguments and annotations overview.

Update example drainOngoingCallTimeout values (60s, 1m, 35s, 30s) to 5s across the runtime config page, actors API reference, and .NET and JavaScript SDK docs, so the examples are no longer silently clamped and stay within the Placement dissemination budget. Add the 2 second default to the API reference tables.