fix(linear): add bias once, not once per SIMD lane

acc was seeded with @splat(bias) and reduced with .Add, so every
linear output carried 7 extra copies of its bias. conv2d was unaffected
(bias via memset), which is why the trunk matched zig-solver to 5e-6
while fc/head activations diverged — the 81% labels regression.

Also adds a scalar tail for in_n < 8 (pose_fc has in_n=4) and a
known-answer selftest (test_linear).

👾 Generated with [Letta Code](https://letta.com)

Co-Authored-By: Letta Code <noreply@letta.com>

build.zig

@@ -5,7 +5,7 @@ pub fn build(b: *std.Build) void {
     const optimize = b.standardOptimizeOption(.{});
 
     // The library module — what downstream consumers import.
-    const crucible_mod = b.addModule("crucible", .{
+    const cruc_mod = b.addModule("crucible", .{
         .root_source_file = b.path("src/crucible.zig"),
         .target = target,
         .optimize = optimize,
@@ -14,19 +14,18 @@ pub fn build(b: *std.Build) void {
     // Static lib artifact (optional install).
     const lib = b.addLibrary(.{
         .name = "crucible",
-        .root_module = crucible_mod,
+        .root_module = cruc_mod,
     });
     b.installArtifact(lib);
 
-    // Example executable: stripsolver (needs a tests/ png decoder? no —
-    // example is self-contained: it uses crucible only for state dict +
-    // layers; arm input comes from CSV).
+    // Example executable: stripsolver (self-contained: uses crucible only
+    // for state dict + layers; arm input comes from CSV).
     const example_mod = b.createModule(.{
         .root_source_file = b.path("examples/stripsolver.zig"),
         .target = target,
         .optimize = optimize,
         .imports = &.{
-            .{ .name = "crucible", .module = crucible_mod },
+            .{ .name = "crucible", .module = cruc_mod },
         },
     });
     const example = b.addExecutable(.{
@@ -41,8 +40,20 @@ pub fn build(b: *std.Build) void {
     const run_step = b.step("run", "Run the stripsolver example");
     run_step.dependOn(&run_example.step);
 
+    // linear selftest
+    const tl_mod = b.createModule(.{
+        .root_source_file = b.path("examples/test_linear.zig"),
+        .target = target,
+        .optimize = optimize,
+        .imports = &.{
+            .{ .name = "crucible", .module = cruc_mod },
+        },
+    });
+    const tl = b.addExecutable(.{ .name = "test_linear", .root_module = tl_mod });
+    b.installArtifact(tl);
+
     // Unit tests.
-    const tests = b.addTest(.{ .root_module = crucible_mod });
+    const tests = b.addTest(.{ .root_module = cruc_mod });
     const run_tests = b.addRunArtifact(tests);
     const test_step = b.step("test", "Run unit tests");
     test_step.dependOn(&run_tests.step);