fix: keep high concurrency Qwen3-TTS unseeded by default

Signed-off-by: Sy03 <1370724210@qq.com>

vllm_omni/deploy/qwen3_tts_high_concurrency.yaml

@@ -60,7 +60,6 @@ stages:
       temperature: 0.9
       top_k: 50
       max_tokens: 4096
-      seed: 42
       repetition_penalty: 1.05
     subtalker_sampling_params:
       do_sample: true
@@ -85,7 +84,6 @@ stages:
       top_p: 1.0
       top_k: -1
       max_tokens: 65536
-      seed: 42
       repetition_penalty: 1.0
 
 platforms: