selfhost/cmd/w6a/parse + lib/ww/lex: amalloc → alloc([], N)! (α-8)

Phase 0 #8 eighth α-batch. 3 sites:
 - w6a/parse.ww:198 nextline newline-hit branch
 - w6a/parse.ww:208 nextline EOF-no-newline branch
 - lib/ww/lex/lex.ww:457 float underscore-strip buffer

All α: alloc([], n+1)! + p[k] indexing + .ptr at the consumer
(`return buf.ptr, n` for parse; `parsef64(clean.ptr, j)` for lex).
No struct-field shape change, no escape.

Verified 132/132 + 995_self_rebuild byte-identity.

lib/ww/lex/lex.ww

@@ -454,7 +454,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 		// Strip underscores from the digits (Hare allows 1_000.5)
 		// before parsing — match what cmd/wcc/lex.c does with
 		// strtod over a cleaned buffer.
-		let clean: *u8 = amalloc(l.a, n + 1u64): *u8;
+		let clean: []u8 = alloc([], n + 1u64)!;
 		let i: u64 = 0u64;
 		let j: u64 = 0u64;
 		for (i < n) {
@@ -466,7 +466,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 			i += 1u64;
 		};
 		clean[j] = 0u8;
-		let fv: f64 = parsef64(clean, j);
+		let fv: f64 = parsef64(clean.ptr, j);
 		out.fval = fv;
 		// Stash the IEEE bits in uval — cgen consumers read floats
 		// as integers (n.uval) to avoid an SSE round-trip when

selfhost/cmd/w6a/main.combined.ww

@@ -1347,21 +1347,21 @@ fn nextline(a: *asm_) (*u8, u64) = {
 		else { a.pos += 1u64; continue; };
 		// hit newline
 		let n: u64 = a.pos - start;
-		let buf: *u8 = amalloc(a.a, n + 1u64): *u8;
+		let buf: []u8 = alloc([], n + 1u64)!;
 		let i: u64 = 0u64;
 		for (i < n) { buf[i] = a.src[start + i]; i += 1u64; };
 		buf[n] = 0u8;
 		a.pos += 1u64;	// skip newline
-		return buf, n;
+		return buf.ptr, n;
 	};
 	// EOF without trailing newline
 	let n: u64 = a.pos - start;
 	if (n == 0u64) { return nil, 0u64; };
-	let buf: *u8 = amalloc(a.a, n + 1u64): *u8;
+	let buf: []u8 = alloc([], n + 1u64)!;
 	let i: u64 = 0u64;
 	for (i < n) { buf[i] = a.src[start + i]; i += 1u64; };
 	buf[n] = 0u8;
-	return buf, n;
+	return buf.ptr, n;
 };
 
 fn skipws(p: *u8, off: u64, n: u64) u64 = {

selfhost/cmd/w6a/parse.ww

@@ -195,21 +195,21 @@ fn nextline(a: *asm_) (*u8, u64) = {
 		else { a.pos += 1u64; continue; };
 		// hit newline
 		let n: u64 = a.pos - start;
-		let buf: *u8 = amalloc(a.a, n + 1u64): *u8;
+		let buf: []u8 = alloc([], n + 1u64)!;
 		let i: u64 = 0u64;
 		for (i < n) { buf[i] = a.src[start + i]; i += 1u64; };
 		buf[n] = 0u8;
 		a.pos += 1u64;	// skip newline
-		return buf, n;
+		return buf.ptr, n;
 	};
 	// EOF without trailing newline
 	let n: u64 = a.pos - start;
 	if (n == 0u64) { return nil, 0u64; };
-	let buf: *u8 = amalloc(a.a, n + 1u64): *u8;
+	let buf: []u8 = alloc([], n + 1u64)!;
 	let i: u64 = 0u64;
 	for (i < n) { buf[i] = a.src[start + i]; i += 1u64; };
 	buf[n] = 0u8;
-	return buf, n;
+	return buf.ptr, n;
 };
 
 fn skipws(p: *u8, off: u64, n: u64) u64 = {

selfhost/cmd/w6c/main.combined.ww

@@ -4166,7 +4166,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 		// Strip underscores from the digits (Hare allows 1_000.5)
 		// before parsing — match what cmd/wcc/lex.c does with
 		// strtod over a cleaned buffer.
-		let clean: *u8 = amalloc(l.a, n + 1u64): *u8;
+		let clean: []u8 = alloc([], n + 1u64)!;
 		let i: u64 = 0u64;
 		let j: u64 = 0u64;
 		for (i < n) {
@@ -4178,7 +4178,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 			i += 1u64;
 		};
 		clean[j] = 0u8;
-		let fv: f64 = parsef64(clean, j);
+		let fv: f64 = parsef64(clean.ptr, j);
 		out.fval = fv;
 		// Stash the IEEE bits in uval — cgen consumers read floats
 		// as integers (n.uval) to avoid an SSE round-trip when

selfhost/cmd/wwdump/main.combined.ww

@@ -4166,7 +4166,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 		// Strip underscores from the digits (Hare allows 1_000.5)
 		// before parsing — match what cmd/wcc/lex.c does with
 		// strtod over a cleaned buffer.
-		let clean: *u8 = amalloc(l.a, n + 1u64): *u8;
+		let clean: []u8 = alloc([], n + 1u64)!;
 		let i: u64 = 0u64;
 		let j: u64 = 0u64;
 		for (i < n) {
@@ -4178,7 +4178,7 @@ fn lexnum(l: *lex, start: *pos, out: *tok) void = {
 			i += 1u64;
 		};
 		clean[j] = 0u8;
-		let fv: f64 = parsef64(clean, j);
+		let fv: f64 = parsef64(clean.ptr, j);
 		out.fval = fv;
 		// Stash the IEEE bits in uval — cgen consumers read floats
 		// as integers (n.uval) to avoid an SSE round-trip when