Clarify comment re `fixnum?' non-use at the syntax level, and add a note

to the `fixnum?' documentation.

collects/scribblings/reference/numbers.scrbl

@@ -149,7 +149,10 @@ Returns @racket[(and (real? v) (inexact? v))].}
 @defproc[(fixnum? [v any/c]) boolean?]{
 
 Return @racket[#t] if @racket[v] is a @techlink{fixnum}, @racket[#f]
-otherwise.}
+otherwise.
+
+Note: the result of this function is platform-dependent, so using it in
+syntax transformers can lead to platform-dependent bytecode files.}
 
 
 @defproc[(flonum? [v any/c]) boolean?]{

collects/typed-scheme/typecheck/tc-expr-unit.rkt

@@ -21,9 +21,10 @@
 (import tc-if^ tc-lambda^ tc-app^ tc-let^ check-subforms^)
 (export tc-expr^)
 
-;; Is the number a fixnum on all the platforms Racket supports?
-;; This relies on Racket being compiled only on 32+ bit systems.
-;; This check is done at compile time to typecheck literals.
+;; Is the number a fixnum on *all* the platforms Racket supports?  This
+;; works because Racket compiles only on 32+ bit systems.  This check is
+;; done at compile time to typecheck literals -- so use it instead of
+;; `fixnum?' to avoid creating platform-dependent .zo files.
 (define (portable-fixnum? n)
   (and (exact-integer? n)
        (< n (expt 2 31))