feat(alpha): alpha_linear_q kernels + launchers for Task 12 DQN smoke

Phase E.1 Task 12a. Three new CUDA kernels for the H=600 DQN smoke
(Task 12 proper) that lands in a follow-up commit:

  alpha_linear_q_forward_kernel    Q = X · W^T + b
  alpha_linear_q_grad_kernel       dW, db sparse MSE-TD over taken actions
  alpha_linear_q_sgd_step_kernel   element-wise params -= lr · grad

Architecture: single linear layer, no hidden layer. The Phase E state
vector has meaningful direct features (alpha_logit, spread_bps, position,
ofi_sum_5, …) so linear Q can capture real relations like Q[Buy] ∝
alpha_logit. If linear can't pass the kill-criteria gate, no architecture
upgrade will save it — and the smoke proceeds with NoisyNet escalation
per the plan.

Sparse gradient: only the taken action contributes (standard DQN TD
loss). No atomicAdd needed — one thread per (i, j) loops over the batch
and adds only when actions[b] == i.

GPU contract:
  - No host branches inside any kernel (graph-capture compatible)
  - No atomicAdd (per feedback_no_atomicadd)
  - All compute on GPU (forward, grad, weight update)
  - Tiny launch overhead — fits per-step (batch=64 forward = 576 threads,
    1 block; grad = 99 threads, 1 block)

Three pub(crate) Rust launchers in alpha_kernels.rs match the
launch_apply_pearls pattern. Cubin embedded via include_bytes!.

Smoke test `linear_q_forward_grad_sgd_round_trip_matches_hand_math`
exercises all three kernels end-to-end on a small (batch=2, state_dim=2,
n_actions=3) case with full hand-math:

  Forward:  Q = [[2.1, 3.2, 0.3], [4.1, 5.2, 0.3]] ✓
  Grad:     dW = [[-1.8, -2.7], [0.8, 1.0], [0, 0]]
            db = [-0.9, 0.2, 0] ✓
  SGD:      W' = [[1.18, 0.27], [-0.08, 0.90], [0, 0]]
            b' = [0.19, 0.18, 0.30] ✓

All within 1e-4 tolerance. `cargo test -p ml --lib alpha_kernels`:
5/5 pass on RTX 3050 Ti in 2.04s (compile witness + 4 GPU smokes).
Audit doc docs/isv-slots.md updated per Invariant 7.

crates/ml/build.rs

@@ -1601,6 +1601,13 @@ fn main() {
         // plus a soft-V bootstrap V_soft(s') = τ · log Σ exp(Q/τ).
         // One thread per batch sample; log-sum-exp-stable softmax.
         "alpha_munchausen_target.cu",
+        // Phase E.1 Task 12a (2026-05-15): linear Q-network kernels for
+        // the H=600 DQN smoke. Forward (Q = X·W^T + b), MSE-TD gradient
+        // (sparse over actions), and plain-SGD step. No hidden layer —
+        // smoke validates whether ANY signal propagates; if linear Q
+        // can't learn the kill-criteria gate, no architecture upgrade
+        // will save it.
+        "alpha_linear_q.cu",
     ];
 
     // ALL kernels get common header (BF16 types + wrappers)