Make zfs-share service resilient to stale exports

The are a few cases where stale entries in /etc/exports.d/zfs.exports 
will cause the nfs-server service to fail when starting up.

Since the nfs-server startup consumes /etc/exports.d/zfs.exports, the 
zfs-share service (which rebuilds the list of zfs exports) should run 
before the nfs-server service.

To make the zfs-share service resilient to stale exports, this change 
truncates the zfs config file as part of the zfs share -a operation.

Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Don Brady <don.brady@delphix.com>
Closes #13775

cmd/zfs/zfs_main.c

@@ -7093,6 +7093,9 @@ share_mount(int op, int argc, char **argv)
 		share_mount_state.sm_total = cb.cb_used;
 		pthread_mutex_init(&share_mount_state.sm_lock, NULL);
 
+		/* For a 'zfs share -a' operation start with a clean slate. */
+		zfs_truncate_shares(NULL);
+
 		/*
 		 * libshare isn't mt-safe, so only do the operation in parallel
 		 * if we're mounting. Additionally, the key-loading option must