[dashboard] fix: accept integer vram status values

AGENTS.md

@@ -342,6 +342,8 @@ This file defines how coding agents should work in this repository.
 - Dashboard refresh must treat malformed successful `/api/gpus` and
   `/api/sessions` list envelopes or records as refresh warnings without
   replacing the last known good dashboard state with empty or malformed data.
+  Session refresh validation must match the public session contract, including
+  string or integer-byte `params.vram` values.
 - Dashboard stop requests must use a timeout that covers the backend startup
   stop wait plus release timeout and a small client buffer; shorter dashboard
   refresh/start requests should keep their modest timeout.

docs/guides/mcp.md

@@ -158,6 +158,8 @@ seconds, including fractional seconds, within the Python runtime wait limit;
 `vram` accepts human-readable sizes or bytes. Byte-equivalent requests below
 4 bytes or above 1 PiB are rejected as public validation errors, and internal
 tensor element counts round up to cover the requested byte-equivalent amount.
+Dashboard session refreshes preserve both string and integer-byte `vram` values
+from valid service status records.
 Cheap local fields (`vram`, `interval`, `busy_threshold`, `job_id`, duplicate
 custom `job_id`, and `gpu_ids` shape) are rejected before `/api/sessions` asks
 the service to list visible GPUs, so bad requests do not spend telemetry work.