Improve SDPA to handle GQA and update models to use native GQA

Add GQA/MQA support to the Triton SDPA kernel with a "pack GQA"
optimization adapted from FlashAttention. When enable_gqa=True and
H_q > H_kv, the kernel folds multiple Q heads sharing the same KV
head into the M (sequence) dimension, so K/V are loaded once per KV
head instead of once per Q head. A tile-utilization heuristic from
FlashAttention decides when packing is beneficial (decode) vs when
simple head remapping suffices (prefill).

Update Qwen 3.5 MoE and Voxtral Realtime models to use native
enable_gqa=True instead of manually expanding KV heads via
repeat_interleave, eliminating redundant memory traffic.

Author: mergennachin