where mv_inner.timestamp > '2020-11-06'::timestamp - interval '1 day' and mv_inner.timestamp < '2020-11-07'::timestamp + interval '1 day'
offset 0
==>
Seq Scan on measurement_value mv_inner (cost=0.00..7175777.00 rows=1219616 width=1006) Filter: (("timestamp" > '2020-11-05 00:00:00'::timestamp without time zone) AND ("timestamp" < '2020-11-08 00:00:00'::timestamp without time zone))
Curious, how accurate is that row count of 1.2 million records for 3 days? How many total records in the table? If you disable sequential scan, does it choose the index and what cost does it show?