Should guard-foreign-escape use `#:callback-exns?`?

[LiberalArtist]

Now that the #:callback-exns? option for _fun has been added (in racket/racket@bf8741e), maybe guard-foreign-escape should use it:

draw/draw-lib/racket/draw/unsafe/callback.rkt

Lines 7 to 18 in 253dbf8

	;; The 'racket VM can handle concurrent callbacks in different Racket
	;; threads, because it copies the C stack in and out to implement
	;; threads. The 'chez-scheme VM cannot do that, so callbacks have to
	;; be atomic. Additional work is needed to allow erroe escapes; we
	;; assume that any error happens in an atomic callback that was
	;; triggered by a foreign call that disables interrupts (which seems
	;; like a fragile assumption!).
	;;
	;; Atomicity implies that a callback cannot read from or write to an
	;; arbitrary port, so we have to "sanitize" a port by adding an
	;; intermediary. Unfortunately, this means that reading from a port
	;; has to be eager.

[mflatt]

Hm... I thought it was, but that's because I've had the change (uncommitted, as it turns out) in my checkout for months!

I'd be much more comfortable if we had been testing this change for 6 weeks, instead of making it just before a release. So, I'll probably wait until after the release branch before merging.