Make faster ProcessFeedService

[harukasan]

In our slow queries I found:

SELECT "accounts".* FROM "accounts" WHERE "accounts"."uri" = '...' LIMIT 1;

and the query plan is:

paon=> EXPLAIN ANALYZE SELECT "accounts".* FROM "accounts" WHERE "accounts"."uri" = '...' LIMIT 1;
 Limit  (cost=0.00..7710.39 rows=1 width=899) (actual time=202.203..202.203 rows=0 loops=1)
   ->  Seq Scan on accounts  (cost=0.00..46262.32 rows=6 width=899) (actual time=202.201..202.201 rows=0 loops=1)
         Filter: ((uri)::text = '...'::text)
         Rows Removed by Filter: 297178
 Planning time: 1.129 ms
 Execution time: 202.238 ms
(6 rows)

It is due to "accounts"."uri" have no index.

[alpaca-tc]

Please remove blank line