comptime compiler-API: Phase 1 foundation + Phase 2.1 weld plan

Introduce the welded comptime `compiler` library (`#library "compiler"` +
`abi(.zig) extern compiler`), per design/comptime-compiler-api.md, and unify
`callconv(...)` into the new `abi(...)` annotation.

abi(...) replaces callconv(...):
- New ABI enum { default, c, zig, pure }; `abi(.c|.zig|.pure)` parses in the
  postfix slot before extern/export (and standalone). `kw_callconv` -> `kw_abi`.
- Migrated 52 sx files, the call-convention-mismatch diagnostic, and docs
  (readme/specs) from `callconv(.c)` to `abi(.c)`.

Phase 1 — welded compiler library (parse -> registry -> validation -> bridge):
- `abi(.zig) extern compiler` parses on fn decls (carries abi/extern_lib) and
  struct decls (StructDecl.abi/extern_lib).
- `#library "compiler"` is the comptime-only internal surface — never dlopen'd.
- src/ir/compiler_lib.zig: the binding registry (the safety boundary). `Field`
  welded to StructInfo.Field with layout baked from the real Zig type
  (@offsetOf/@sizeOf); `findType`/`findFn`. Welded structs are layout-validated
  at registration (field set + total size) as a header checked against the impl.
- Host-call bridge: a `fn abi(.zig) extern compiler` dispatches under the
  comptime interp to its registered Zig handler (intern/text_of round-trip),
  never dlsym. IR Function.compiler_welded; validated in declareFunction.
- Comptime-only enforcement: a runtime call to a welded fn is a clean
  build-gating error (emitCall), not an undefined-symbol link failure.

Phase 2.1 — byte-layout weld foundation:
- Decision: full byte-layout weld (sx struct laid out byte-identically to the
  bound Zig type). Registered StructInfo (first non-natural / Zig-reordered
  layout). `computeWeldPlan` — pure offset-ordered element plan + padding +
  sx-field->LLVM-element remap; unit-tested. Emit/interp wiring is the next
  sub-step (2.2+, see current/CHECKPOINT-COMPILER-API.md).

Examples: 0625/0626 (welded struct + fn round-trip), 1183/1184/1185
(layout-mismatch, unexported-fn, runtime-call diagnostics).

library/modules/platform/android.sx

@@ -85,7 +85,7 @@ ANativeWindow_setBuffersGeometry :: (w: *void, width: i32, height: i32, fmt: i32
 AAssetManager_fromJava :: (env: *void, mgr: *void) -> *void extern;
 
 // pthread (link libpthread is built into bionic).
-pthread_create  :: (thread: *u64, attr: *void, start: (*void) -> *void callconv(.c), arg: *void) -> i32 extern;
+pthread_create  :: (thread: *u64, attr: *void, start: (*void) -> *void abi(.c), arg: *void) -> i32 extern;
 pthread_mutex_init    :: (m: *void, attr: *void) -> i32 extern;
 pthread_mutex_lock    :: (m: *void) -> i32 extern;
 pthread_mutex_unlock  :: (m: *void) -> i32 extern;
@@ -254,7 +254,7 @@ sx_android_start_render_thread :: (plat: *AndroidPlatform, entry_fn: () -> void)
     plat.render_thread_started = true;
 }
 
-sx_android_render_thread_entry :: (arg: *void) -> *void callconv(.c) {
+sx_android_render_thread_entry :: (arg: *void) -> *void abi(.c) {
     plat : *AndroidPlatform = xx arg;
     while plat.app_window == null and !plat.should_stop {
         usleep(1000);

library/modules/platform/sdl3.sx

@@ -229,7 +229,7 @@ impl Platform for SdlPlatform {
 // SDL fires the watch synchronously when events are added — including during
 // macOS's modal resize-drag, when SDL_PollEvent can't run. Re-invoking the
 // frame closure here keeps content rendering at the new size during the drag.
-sdl_event_watch :: (userdata: *void, event: *SDL_Event) -> bool callconv(.c) {
+sdl_event_watch :: (userdata: *void, event: *SDL_Event) -> bool abi(.c) {
     plat : *SdlPlatform = xx userdata;
     if event.* == {
         case .window_resized: (data) {
@@ -245,7 +245,7 @@ sdl_event_watch :: (userdata: *void, event: *SDL_Event) -> bool callconv(.c) {
     true
 }
 
-sdl_wasm_tick :: () callconv(.c) {
+sdl_wasm_tick :: () abi(.c) {
     if g_sdl_plat == null { return; }
     if !g_sdl_plat.has_frame_closure { return; }
     fn := g_sdl_plat.frame_closure;

library/modules/platform/uikit.sx

@@ -790,8 +790,8 @@ impl Platform for UIKitPlatform {
 }
 
 // dlsym(RTLD_DEFAULT, name) — Apple platforms. RTLD_DEFAULT is (void*)-2.
-// callconv(.c) so this is callable from `load_gl`'s C-conv proc-loader slot.
-ios_gl_proc :: (name: [*]u8) -> *void callconv(.c) {
+// abi(.c) so this is callable from `load_gl`'s C-conv proc-loader slot.
+ios_gl_proc :: (name: [*]u8) -> *void abi(.c) {
     rtld_default : *void = xx (0 - 2);
     dlsym(rtld_default, name)
 }