Fix get replication task read level update issue

[Shaddoll]

What changed?
Update replication read level for skipped tasks

Why?
If the replication tasks read in a batch belong to a domain which isn't replicated to the polling cluster, the read level won't be updated and the polling cluster will try to fetch the same replication tasks and the replication of polling cluster will get stuck and the replication tasks from the source cluster won't be cleaned up.
We didn't hit this issue because we only have 2 replicas, and that branch is never hit in this situation.

How did you test it?
existing test

Potential risks

Release notes

Documentation Changes

[coveralls]

Pull Request Test Coverage Report for Build 0426e737-0318-4882-a888-489330aba031

• 1 of 2 (50.0%) changed or added relevant lines in 1 file are covered.
• 16 unchanged lines in 6 files lost coverage.
• Overall coverage decreased (-0.005%) to 56.922%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
service/history/replication/task_ack_manager.go	1	2	50.0%

Files with Coverage Reduction	New Missed Lines	%
common/task/weightedRoundRobinTaskScheduler.go	1	89.64%
service/history/replication/task_ack_manager.go	1	85.77%
common/types/mapper/thrift/shared.go	2	63.43%
common/cache/lru.go	3	90.73%
common/task/fifoTaskScheduler.go	3	84.54%
service/history/execution/context.go	6	73.96%

Totals
Change from base Build 637a52ef-723a-4719-8238-780b58dbfcb4:	-0.005%
Covered Lines:	82016
Relevant Lines:	144085

---

💛 - Coveralls