ZTS: Fix enospc_002_pos.ksh again

This is a follow up commit for e03a41a60 which aimed to resolve
this same test failure.  The core "problem" here is that it takes
very little space to perform a clone/snapshot/bookmark, which
means if we want these commands to reliably fail the pool must
truely have exhausted all free space.

This commit increases the number of fill iterations to try and
consume every block which we can.  This still can't guarantee
the clone/snapshot/bookmark will fail, but it significantly
improves the odds.  The exception was kept since it's still
not a sure thing.

Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #12903

tests/zfs-tests/tests/functional/no_space/enospc_002_pos.ksh

@@ -49,7 +49,7 @@ log_must zfs snapshot $TESTPOOL/$TESTFS@snap
 #
 log_note "Writing files until ENOSPC."
 
-for i in $(seq 30); do
+for i in $(seq 100); do
 	file_write -o create -f $TESTDIR/file.$i -b $BLOCKSZ \
 	    -c $NUM_WRITES -d $DATA
 	ret=$?