refactor(cuda): eliminate all CPU fallbacks — CUDA mandatory across ML stack

- Remove ALL #[cfg(feature = "cuda")] guards (~400+ occurrences)
- Remove ALL #[cfg_attr(not(feature = "cuda"), ignore)] test annotations (~250)
- Make cuda default feature in 9 ML crates (ml, ml-core, ml-dqn, ml-ppo, etc.)
- Convert nvrtc JIT compilation to precompiled nvcc (searchsorted, prefix_sum)
- Move compile_ptx_for_device() to ml-core for shared access
- Delete dead CPU code: multi_step.rs, self_supervised_pretraining.rs,
  training_guard_gpu_tests.rs, CPU PER buffer paths, CPU Q-diagnostics
- Replace unwrap_or(Device::Cpu) with hard errors everywhere
- Remove dead is_cuda() else branches in DQN/PPO/hyperopt trainers
- Change config defaults from "cpu" to "cuda" (rainbow, tlob, pipeline)
- Port IQL value network to GPU kernel (5 CUDA entry points)
- Port HER goal relabeling to GPU kernel (warp-per-sample)
- Wire DSR GPU-to-CPU sync in training loop
- cfg!(feature = "cuda") → true in inference_validator

Zero warnings, zero errors across entire workspace.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

crates/ml/Cargo.toml

@@ -13,9 +13,9 @@ keywords.workspace = true
 categories.workspace = true
 
 [features]
-# MINIMAL features for HFT inference only - ALL HEAVY ML REMOVED
-# CUDA opt-in: service crates build on CPU nodes without nvcc
-default = ["minimal-inference"]
+# CUDA default: all ML training is GPU-only. Service crates on CPU nodes
+# must opt out with `default-features = false, features = ["minimal-inference"]`.
+default = ["minimal-inference", "cuda"]
 
 # PRODUCTION FEATURES - LIGHTWEIGHT ONLY
 minimal-inference = []  # Minimal inference with no optional deps