config: the last runner-side instance reads read the bags

Six reads were left on `self.server_args` outside the per-instance boundary the
plan reserves for the tokenizer-manager family, and each had a different reason
to be there:

- `scheduler.process_input_requests` (`mm_feature_transport`) and
  `BaseSpecWorker._build_hicache_draft_plan` (`enable_hierarchical_cache`) are
  plain leaves -> `get_mm()` / `get_memory()`.
- `DraftBackendFactory._create_backend` read the split backend through a
  *runtime-computed name* (`getattr(self.server_args, backend_name)`) and then
  fell back to the base field by hand -- the census's documented blind spot.
  The two names it can be handed are exactly the pair `attention_backends()`
  returns with that fallback already applied, so it reads the pair and indexes
  it. The draft runner's own stamp still wins when it has one.
- `remote_instance_weight_loader_use_transfer_engine` and
  `pre_capture_activation_reserve_mb` are derived members. Both are computed
  from published leaves only, so both get a named accessor that derives from
  the bags (and therefore follows a post-publish override).

The first of those two has all its inputs in one bag, so it follows the
established shape: one `*_of(cfg)` helper in `arg_groups/overrides.py`, the
`ServerArgs` member delegating to it, and the accessor calling it on
`get_model()`. `modelexpress_transport_of` splits out the JSON parse both
sides need. The second spans four bags plus the configured parallel sizes, so
it exists twice like the mamba pair -- and `TestDerivedPredicatesAgreeAcrossTiers`
now pins both new pairs equal over their input matrices (92 subtests).

`self.server_args.X` outside the tokenizer-manager family: 11 -> 5, and the
five that remain are the documented ones (the encode server's own record, the
nixl connector's rank arithmetic, `GrammarManager`'s handed instance).

The post-capture headroom path calls the same bag-backed
`pre_capture_activation_reserve_mb` accessor the configurator uses -- the
accessor advertises override-following, and a reserve that reads the record
while its sibling reads the bags can disagree after a post-publish override.
And the conversions' orphans go with them: `RemoteInstanceWeightTransporter`
kept a `server_args` field nothing reads, and `DraftBackendFactory` parked a
record it no longer consults -- both drop the parameter, and the four factory
call sites stop threading one.

python/sglang/srt/server_args.py

@@ -44,6 +44,7 @@ from sglang.srt.arg_groups.overrides import (
     attention_backends_of,
     mamba_extra_buffer_lazy_of,
     mamba_extra_buffer_of,
+    remote_instance_transfer_engine_of,
     resolved_view,
 )
 from sglang.srt.configs.embedding_model_spec import BCGPrefillPolicy
@@ -9470,20 +9471,7 @@ class ServerArgs:
     def remote_instance_weight_loader_use_transfer_engine(self, load_format=None):
         """``load_format`` overrides the seed's: a draft runner loading under
         ``--speculative-draft-load-format`` needs its own transfer engine."""
-        # Use TransferEngine as seed backend.
-        if self.remote_instance_weight_loader_start_seed_via_transfer_engine:
-            return True
-        # Use TransferEngine as client backend.
-        if (load_format or self.load_format) == "remote_instance" and (
-            self.remote_instance_weight_loader_backend == "transfer_engine"
-            or (
-                self.remote_instance_weight_loader_backend == "modelexpress"
-                and self.modelexpress_transport == "transfer_engine"
-            )
-        ):
-            return True
-        else:
-            return False
+        return remote_instance_transfer_engine_of(self, load_format)
 
     def describe_kv_events_publisher(self) -> Optional[dict]:
         """Return a structured description of this server's KV-event