help interpreting pg_stat_user_index view values

Need some help interpreting the results of queries against the
pg_stat_user_index view.

Given the following four contrieved indexes and their scan, read and fetch
values in pg_stat_user_index view:

Index name   idx_scan   idx_tup_read   idx_tup_fetch
        idx1               100                    0
0
        idx2               100                  200                        0
        idx3               100                  200                       50
        idx4               100                    0
200

Is idx1 a "useless" index?  Is it being scanned but nevering returns useful
tuples because it doesn't point to any useful rows in the table? Or maybe
the query planner looked at the index but decided to use a table scan
instead?

Is idx2 a "useless" index?  Is this index being scanned but nevering returns
useful tuples because it doesn't point to any useful rows in the table?

For idx3 do it's values mean it's column specificity is not specific enough
to be a relatively useful index?

I am assuming an index with values like idx4 could never exist, it is an
impossible result.  Is that a correct assumption?