scripts: Disentangled -r/--hot and -i/--enumerate

This removes most of the special behavior around how -r/--hot and
-i/--enumerate interact. This does mean -r/--hot risks folding results
if -i/--enumerate is not specified, but this is _technically_ a valid
operation.

For most of the recursive result scripts, I've replaced the "i" field
with separate "z" and "i" fields for depth and field number, which I
think is a bit more informative/useful.

I've also added a default-hidden "off" field to structs.py/ctx.py, since
we have that info available. I considered replacing "i" with this, but
decided against it since non-zero offsets for union members would risk
being confusing/mistake prone.

scripts/code.py

@@ -922,11 +922,13 @@ def write_csv(path, Result, results, *,
     with openio(path, 'w') as f:
         # write csv?
         if not json:
-            writer = csv.DictWriter(f,
-                    (by if by is not None else Result._by)
-                        + [k for k in (fields
-                            if fields is not None
-                            else Result._fields)])
+            writer = csv.DictWriter(f, list(co.OrderedDict.fromkeys(it.chain(
+                    by
+                        if by is not None
+                        else Result._by,
+                    fields
+                        if fields is not None
+                        else Result._fields)).keys()))
             writer.writeheader()
             for r in results:
                 # note this allows by/fields to overlap