fix: xx pack[i] to a protocol target heap-copies the element

Erasing a single comptime-pack element to a protocol value
(`xx sources[0]` with a protocol target) tripped the pack-as-value
error: buildProtocolErasure treated the index_expr as an lvalue and
took its address via lowerExprAsPtr, whose .index_expr arm lowers the
bare pack as a value (a pack is comptime-only with no runtime storage).

isLvalueExpr now reports a comptime pack index as an rvalue, decided
via the same packArgNodeAt predicate the value path uses — so the value
and lvalue paths can't diverge on what counts as a pack element — and
erasure heap-copies the already-materialized element instead.

Resolves issue 0135. Regression tests: examples/0547, 0548.

examples/0547-packs-xx-pack-index-to-protocol.sx

@@ -0,0 +1,27 @@
+// `xx <pack>[i]` erased to a protocol-typed local.
+//
+// Erasing a single comptime-pack element to a protocol scalar routes through
+// buildProtocolErasure. A pack index is a comptime rvalue (a pack has no
+// runtime storage — `sources[i]` resolves to the call-site arg, which only
+// gets storage when lowered as a value), so the erasure must heap-copy the
+// materialized element rather than take its address.
+//
+// Regression (issue 0135): `xx sources[0]` used to lower the bare pack as a
+// value and error with "pack 'sources' has no runtime value".
+
+#import "modules/std.sx";
+
+VL :: protocol(T: Type) { get :: () -> T; }
+IntCell :: struct { v: i64; }
+impl VL(i64) for IntCell { get :: (self: *IntCell) -> i64 => self.v; }
+
+first :: (..sources: VL) -> i64 {
+    x : VL(i64) = xx sources[0];   // erase element 0 to VL(i64)
+    return x.get();
+}
+
+main :: () -> i32 {
+    print("{}\n", first(IntCell.{ v = 7 }));                       // 7
+    print("{}\n", first(IntCell.{ v = 42 }, IntCell.{ v = 99 }));  // 42 (element 0)
+    0
+}

examples/0548-packs-xx-pack-index-two-elements.sx

@@ -0,0 +1,25 @@
+// Erase two DISTINCT comptime-pack elements to protocol locals — each gets
+// its own heap copy and resolves to its OWN concrete type's method (IntCell.get
+// vs Doubler.get), proving the per-element erasure picks the right vtable.
+//
+// Regression (issue 0135): single-element `xx pack[i]` erasure to a protocol
+// scalar was unsupported (the bare pack lowered as a value and errored).
+
+#import "modules/std.sx";
+
+VL :: protocol(T: Type) { get :: () -> T; }
+IntCell :: struct { v: i64; }
+impl VL(i64) for IntCell { get :: (self: *IntCell) -> i64 => self.v; }
+Doubler :: struct { n: i64; }
+impl VL(i64) for Doubler { get :: (self: *Doubler) -> i64 => self.n * 2; }
+
+sum_two :: (..sources: VL) -> i64 {
+    a : VL(i64) = xx sources[0];   // erase element 0
+    b : VL(i64) = xx sources[1];   // erase element 1
+    return a.get() + b.get();
+}
+
+main :: () -> i32 {
+    print("{}\n", sum_two(IntCell.{ v = 10 }, Doubler.{ n = 16 }));  // 10 + (16*2) = 42
+    0
+}

examples/expected/0547-packs-xx-pack-index-to-protocol.exit

@@ -0,0 +1 @@
+0

examples/expected/0547-packs-xx-pack-index-to-protocol.stderr

@@ -0,0 +1 @@
+

examples/expected/0547-packs-xx-pack-index-to-protocol.stdout

@@ -0,0 +1,2 @@
+7
+42

examples/expected/0548-packs-xx-pack-index-two-elements.exit

@@ -0,0 +1 @@
+0

examples/expected/0548-packs-xx-pack-index-two-elements.stderr

@@ -0,0 +1 @@
+

examples/expected/0548-packs-xx-pack-index-two-elements.stdout

@@ -0,0 +1 @@
+42