fix byte 'downto 0' for-loop hanging on the virtual target

BSTPOS after DECM read stale status flags on the VM because DECM leaves
flags untouched there. Only use the BSTPOS fast path on 6502 hardware or
targets honoring the multi-byte status-bits contract; otherwise emit the
explicit LOADM+CMPI+BSTNE check. Backported from master (22a2f8cbc).

codeCore/src/prog8/code/core/ICompilationTarget.kt

@@ -44,6 +44,11 @@ enum class CpuType(val statusBitsOnMultiByteOps: Boolean) {
      * behavior consistent across all current targets.
      */
     VIRTUAL(false)
+    ;
+
+    /** True for the 6502-family targets (plain 6502 and 65C02). */
+    val is6502: Boolean
+        get() = this == CPU6502 || this == CPU65C02
 }
 
 interface ICompilationTarget: IStringEncoding, IMemSizer {

codeGenIntermediate/src/prog8/codegen/intermediate/IRCodeGen.kt

@@ -888,7 +888,10 @@ class IRCodeGen(
         val chunk2 = addConstMem(loopvarDtIr, null, loopvarSymbol, iterable.step)
         if(loopvarDtIr==IRDataType.BYTE && iterable.step==-1 && iterable.last==0) {
             // downto 0 optimization (byte)
-            if(loopvarDt.isSignedByte || iterable.first<=127) {
+            // Only rely on DECM setting status bits on targets where it actually does (6502 hardware,
+            // or targets that honor the strict multi-byte status-bits contract). On the VM DECM
+            // leaves flags untouched, so BSTPOS would read stale state and loop forever.
+            if((loopvarDt.isSignedByte || iterable.first<=127) && (options.compTarget.cpu.is6502 || options.compTarget.cpu.statusBitsOnMultiByteOps)) {
                 chunk2 += IRInstruction(Opcode.BSTPOS, labelSymbol = loopLabel)
             } else {
                 chunk2 += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = indexReg, labelSymbol = loopvarSymbol)

examples/test.p8

@@ -1,14 +1,12 @@
-main {
-    const uword STACK_SIZE = 10
-    struct Stack {
-        uword[STACK_SIZE] stack1
-        uword[20] stack2
-        uword top
-    }
+%import textio
+%zeropage basicsafe
 
+main {
     sub start() {
-        ^^Stack st = []
-        st.stack1[5] = 9999
-        st.stack2[5] = 8888
+
+        for cx16.r0L in 5 downto 0 {
+            txt.print_ub(cx16.r0L)
+            txt.spc()
+        }
     }
 }

gradle.properties

@@ -4,4 +4,4 @@ org.gradle.parallel=true
 org.gradle.daemon=true
 org.gradle.configuration-cache=false
 kotlin.code.style=official
-version=12.3.4-SNAPSHOT
+version=12.3.4