WRITE_HOLD machinery: no need for to bump mount_lock seqcount

WRITE_HOLD machinery: no need for to bump mount_lock seqcount

... neither for insertion into the list of instances, nor for
mnt_{un,}hold_writers(), nor for mnt_get_write_access() deciding
to be nice to RT during a busy-wait loop - all of that only needs
the spinlock side of mount_lock.

IOW, it's mount_locked_reader, not mount_writer.

Clarify the comment re locking rules for mnt_unhold_writers() - it's
not just that mount_lock needs to be held when calling that, it must
have been held all along since the matching mnt_hold_writers().

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>