hexagon: add support for CONCAT op (#23648)

* hexagon: add support for CONCAT with optimized concat_2d_transposed

qwen3.5 models are quite heavy on the CONCAT with large and transposed src1.

* hex-concat: use fastdiv in generic version

* hex-concat: make checks for transposed a bit more readable

* hex-concat: reoder dma ops for better pipelining

* hex-cont/cpy: optimize CPY and CONT ops

The primary change is to avoid scalar divs in the inner loops.
We were calling hvx_copy_uu(... type_size) where type_size is non a constexpr.
This causes runtime divs by that value which is normally just 4 or 2 (f32/f16).

* hex-get-rows: optimize GET_ROWS for large rows

We now use DMA for larger rows and also split them into chunks to improve perf for Qwen3.5 and other models
that do lots of GET_ROWS with huge (2MB+ rows).

Also bump the DMA queue depth now that we can take advantage of it.

* hex-concat: unroll the inner loops of concat_2d

* hex-concat: more updates to concat_2d to improve perf a bit further

* hex-cpy: fixed n_rows per thread checks in the copy ops

* hmx-fa: fix alignment issues while computing dma sizes

* hex-set-rows: add early returns for idle threads

* hvx-rope: minor optimization to replace loops with fastdiv logic

* hex-rope: replace scalar tail processing with HVX

* hex-rope: optimize rope cache init with HVX

Add hvx-utils sin/cos helpers that use an aprox method (similar to rsqrt, inverse, etc)
Use the helpers to optimize ROPE.

ggml/src/ggml-hexagon/htp/unary-ops.c

@@ -207,7 +207,7 @@ static void hvx_fast_norm_f32(const uint8_t * restrict src,
 
     // scale = rsqrt(variance + epsilon),  mean_x broadcast for subtraction
     HVX_Vector scale_v  = hvx_vec_rsqrt_f32(Q6_Vsf_equals_Vqf32(var_epsilon_v));
-    HVX_Vector mean_x_b = hvx_vec_splat_f32(hvx_vec_get_f32(Q6_Vsf_equals_Vqf32(mean_x_v)));
+    HVX_Vector mean_x_b = hvx_vec_repl_f32(Q6_Vsf_equals_Vqf32(mean_x_v));
 
     #pragma unroll(4)
     for (int i = 0; i < nvec; i++) {