IC: next steps (#16632)

* removed dead code
* beginnings of a rodfile reader
* IC: record global VM state changes and pragma state changes
* IC: replay pragmas and VM state changes
* implemented rod load file simuation for easier, extensive testing
* critical bugfix
* IC: stress test logic; should also help with recursive module dependencies; WIP
* IC: loading from .rod files begins to work reliably
* removed ugly hacks
* yet another silly mistake

compiler/ast.nim

@@ -221,6 +221,8 @@ type
     nkBreakState,         # special break statement for easier code generation
     nkFuncDef,            # a func
     nkTupleConstr         # a tuple constructor
+    nkModuleRef           # for .rod file support: A (moduleId, itemId) pair
+    nkReplayAction        # for .rod file support: A replay action
 
   TNodeKinds* = set[TNodeKind]
 
@@ -1139,6 +1141,9 @@ proc add*(father, son: Indexable) =
     if isNil(father.sons): father.sons = @[]
   father.sons.add(son)
 
+proc addAllowNil*(father, son: Indexable) {.inline.} =
+  father.sons.add(son)
+
 template `[]`*(n: Indexable, i: int): Indexable = n.sons[i]
 template `[]=`*(n: Indexable, i: int; x: Indexable) = n.sons[i] = x