I think consistency should have the top priority, and for_list should act like other for loops do.

For-loop over an array will silently skip over deleted elements, hence a for-loop over a hash should too. (no tested yet)

If a warning should be activated, then in both cases alike.

update

unfortunately it's NOT consistent

$ cat tst_del.pl
use v5.14;
use experimental "for_list";

my %h;
@h{"a".."d"} = 1..4;

my @a = values %h;

my $once;


$once = 0;
for my $e (@a) {
  splice @a,0,2 unless $once++;
  say "A: $e";
}


$once = 0;
for my ($k,$v) (%h) {
  delete @h{"a".."b"} unless $once++;
  say "H: $k => $v";
}

say "never reached";

$ perl tst_del.pl 
A: 1
A: 3
H: a => 1
H: d => 4
Use of freed value in iteration at tst_del.pl line 22.

$ perl -v

This is perl 5, version 38, subversion 2 (v5.38.2)
[download]

update

tho, according to from perlsyn it's undefined behavior

    If any part of LIST is an array, "foreach" will get very confused 
+if you
    add or remove elements within the loop body, for example with "spl
+ice".
    So don't do that.
[download]

and the demonstrated results are quite unpredictable.