llama.cpp runtime arm: serve-script runtime switch, harness containerless support, master build

scripts/frank-large.env

@@ -88,6 +88,10 @@ FRANK_LARGE_UTIL=0.45
 FRANK_LARGE_SPEC=
 FRANK_LARGE_SPEC_DIR=/home/dank/models/Qwen3.5-9B
 FRANK_LARGE_SPEC_TOKENS=2
+# AUTORESEARCH: torchao int8 weight-only (quark int8 + all fp8 kernels
+# rejected on gfx1151; torchao is a separate Triton dequant stack)
+FRANK_LARGE_QUANT=torchao
+FRANK_LARGE_QUANT_CONFIG='{"torchao_config":"int8_weight_only"}'
 
 # --- client contract ----------------------------------------------------------
 # vLLM emits the thinking text in `reasoning` (NOT llama.cpp/gufo's