sampling : stop short if backend sampler sampled a token

This commit modifies the graph building logic to immediately continue
when a token has already been sampled by the backend sampler.

It also updates the test for backend temporary sampling to include
top-k and distribution samplers in the chain to verify that they are not
producing any logits (they are not run).

src/llama-graph.cpp

@@ -2100,6 +2100,7 @@ void llm_graph_context::build_sampling() const {
         if (data.sampled != nullptr) {
             res->t_sampled[seq_id] = data.sampled;
             ggml_build_forward_expand(gf, data.sampled);
+            continue;
         }
 
         if (data.probs != nullptr) {