[2/N] Added the core structure of elastic EP and the eplb algorithm with faulty rank (#10606)

Co-authored-by: Xun Sun <UNIDY2002@outlook.com>
Co-authored-by: Shangming Cai <csmthu@gmail.com>

python/sglang/srt/server_args.py

@@ -600,6 +600,9 @@ class ServerArgs:
         # Handle any other necessary validations.
         self._handle_other_validations()
 
+        # Handle elastic expert parallelism.
+        self._handle_elastic_ep()
+
     def _handle_deprecated_args(self):
         # handle deprecated tool call parsers
         deprecated_tool_call_parsers = {"qwen25": "qwen", "glm45": "glm"}
@@ -1225,6 +1228,15 @@ class ServerArgs:
         if self.enable_eplb:
             assert self.ep_size > 1
 
+    def _handle_elastic_ep(self):
+        if self.elastic_ep_backend is not None:
+            if self.enable_eplb:
+                if self.eplb_algorithm == "auto":
+                    self.eplb_algorithm = "elasticity_aware"
+                assert (
+                    self.eplb_algorithm == "elasticity_aware"
+                ), "Elastic EP requires eplb_algorithm to be set to 'auto' or 'elasticity_aware'."
+
     def _handle_expert_distribution_metrics(self):
         if self.enable_expert_distribution_metrics and (
             self.expert_distribution_recorder_mode is None