fix(codegen): QBE stores float to implicit-Self field with stored/stores

Assigning a float value to a class field via an implicit Self reference
(e.g. `V := d` in a method/constructor where V: Double) emitted a hardcoded
`storel` on a `d`/`s` SSA value, which QBE rejects ("invalid type for first
operand in storel"). The implicit-Self field-store path in EmitAssignment
fell through to a literal `storel` instead of StoreInstrFor(field type), and
lacked the d<->s width coercion the EmitFieldAssignment path already has.

Use StoreInstrFor(ISFld.TypeDesc) (-> stored/stores for tyDouble/tySingle) and
add the Single<->Double truncd/exts coercion before the store. Surfaced by the
new float-constructor e2e tests, which run on both backends.

compiler/src/main/pascal/blaise.codegen.qbe.pas

@@ -4056,7 +4056,24 @@ begin
           EmitLine(Format('  call $_ClassAddRef(l %s)',  [ValTemp]));
         EmitLine(Format('  call $_ClassRelease(l %s)', [OldTemp]));
       end;
-      EmitLine(Format('  storel %s, %s', [ValTemp, ObjTemp]));
+      { Float-width coercion before the store: a real-typed RHS lands in the
+        SSA as 'd' even when the field is a 32-bit Single ('s'), and vice
+        versa.  Without this an implicit-Self 'V := d' float-field store
+        emitted 'storel' on a 'd' value — a QBE type error. }
+      if (QType = 's') and (QbeTypeOf(AAssign.Expr.ResolvedType) = 'd') then
+      begin
+        ExtTemp := AllocTemp();
+        EmitLine(Format('  %s =s truncd %s', [ExtTemp, ValTemp]));
+        ValTemp := ExtTemp;
+      end
+      else if (QType = 'd') and (QbeTypeOf(AAssign.Expr.ResolvedType) = 's') then
+      begin
+        ExtTemp := AllocTemp();
+        EmitLine(Format('  %s =d exts %s', [ExtTemp, ValTemp]));
+        ValTemp := ExtTemp;
+      end;
+      EmitLine(Format('  %s %s, %s',
+        [StoreInstrFor(ISFld.TypeDesc), ValTemp, ObjTemp]));
     end;
     Exit;
   end;