Does that mean then that any estimate_percent value other than auto_sample_size (say 25%) will use the old style algorithms?Pretty much. The percentage ends up being pushed down into SAMPLE clauses on the queries that are run to deduce the data distribution. And don't forget that sampling (say) 2% means (by default) 2% of the rows. If you have 50 rows per block, you might well still visit every block.
re: the other comments
Whilst we don't have a tablespace level filter, you could easily set up your own via (pseudocode)
for i in ( select ... from dba_segments where tablespace_name = '...' ) loop
dbms_stats.lock_table_stats
end loop
and then explicitly control the stats on those tables manually.