Surprised by index choice for count(*)

Should I be?  I would have thought the pk would have been chosen v.
function index?

explain analyse select count(*) from bc.segment s;
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
 Finalize Aggregate  (cost=4428009.24..4428009.25 rows=1 width=8)
(actual time=14786.395..14786.395 rows=1 loops=1)
   ->  Gather  (cost=4428009.03..4428009.24 rows=2 width=8) (actual
time=14786.358..14786.386 rows=3 loops=1)
         Workers Planned: 2
         Workers Launched: 2
         ->  Partial Aggregate  (cost=4427009.03..4427009.04 rows=1
width=8) (actual time=14782.167..14782.167 rows=1 loops=3)
               ->  Parallel Index Only Scan using fpv on segment s 
(cost=0.57..4210177.14 rows=86732753 width=0) (actual
time=0.061..11352.855 rows=69386204 loops=3)
                     Heap Fetches: 1780
 Planning time: 0.221 ms
 Execution time: 14815.939 ms
(9 rows)

\d bc.segment
              Table "bc.segment"
     Column     |  Type   |     Modifiers
----------------+---------+--------------------
 id             | uuid    | not null
 chrom          | integer | not null
 markerset_id   | uuid    | not null
 probandset_id  | uuid    | not null
 startbase      | integer | not null
 endbase        | integer | not null
 firstmarker    | integer | not null
 lastmarker     | integer | not null
 events_less    | bigint  | not null default 0
 events_equal   | bigint  | not null default 0
 events_greater | bigint  | not null default 0
Indexes:
    "segment_pkey" PRIMARY KEY, btree (id)
    "useg" UNIQUE, btree (probandset_id, chrom, startbase, endbase)
    "fpv" btree (pv(events_less, events_equal, events_greater, 0))
    "segment_markerset_id_probandset_id_idx" btree (markerset_id,
probandset_id)

create or replace function public.pv(l bigint, e bigint, g bigint, o int)
returns float
as
$$
select 1.0*(g+e+o)/(l+e+g+o)::float;
$$
language sql
;