optimize x % p where p=power-of-2, into just x & (p-1)

compiler/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -490,9 +490,9 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     if(!idt.isKnown)
                         throw FatalAstException("unknown dt")
                     return NumericLiteralValue(idt.typeOrElse(DataType.STRUCT), 0, expr.position)
-                } else if (cv == 2.0) {
+                } else if (cv in powersOfTwo) {
                     expr.operator = "&"
-                    expr.right = NumericLiteralValue.optimalInteger(1, expr.position)
+                    expr.right = NumericLiteralValue.optimalInteger(cv!!.toInt()-1, expr.position)
                     return null
                 }
             }

docs/source/todo.rst

@@ -2,8 +2,6 @@
 TODO
 ====
 
-- optimize %32 to & 31
-
 - optimize several inner loops in gfx2
 - hoist all variable declarations up to the subroutine scope *before* even the constant folding takes place (to avoid undefined symbol errors when referring to a variable from another nested scope in the subroutine)
 - optimize swap of two memread values with index, using the same pointer expression/variable, like swap(@(ptr+1), @(ptr+2))

examples/test.p8

@@ -1,10 +1,17 @@
+%import textio
 
 main {
     sub start() {
         uword width
+        uword width2 = 12345
+        ubyte ub1
+        ubyte ub2 = 123
 
-        ; TODO fix compiler crash with noopt:
-        uword x1 = (width-1 as uword) + 2
+        ub1 = ub2 % 32
+        txt.print_ub(ub1)
+        txt.nl()
+        width = width2 % 32
+        txt.print_uw(width)
 
     }
 }