ZTS: normalize on use of sync_pool and sync_all_pools

- Replaces use of manual `zpool sync`
- Don't use `log_must sync_pool` as `sync_pool` uses it internally
- Replace many (but not all) uses of `sync` with `sync_pool`

This makes the tests more consistent, and makes searching easier.

Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Allan Jude <allan@klarasystems.com>
Closes #12894

tests/zfs-tests/include/blkdev.shlib

@@ -595,7 +595,7 @@ function list_file_blocks # input_file
 	else
 		AWK='awk'
 	fi
-	log_must zpool sync -f
+	sync_all_pools true
 	zdb -dddddd $ds $objnum | $AWK -v pad=$((4<<20)) -v bs=512 '
 	    /^$/ { looking = 0 }
 	    looking {

tests/zfs-tests/include/libtest.shlib

@@ -3104,6 +3104,7 @@ function datasetcksum
 {
 	typeset cksum
 	sync
+	sync_all_pools
 	cksum=$(zdb -vvv $1 | grep "^Dataset $1 \[" | grep "cksum" \
 		| awk -F= '{print $7}')
 	echo $cksum
@@ -3519,6 +3520,24 @@ function sync_pool #pool <force>
 	return 0
 }
 
+#
+# Sync all pools
+#
+# $1 boolean to force uberblock (and config including zpool cache file) update
+#
+function sync_all_pools #<force>
+{
+	typeset force=${1:-false}
+
+	if [[ $force == true ]]; then
+		log_must zpool sync -f
+	else
+		log_must zpool sync
+	fi
+
+	return 0
+}
+
 #
 # Wait for zpool 'freeing' property drops to zero.
 #