changefeed fail to initialize under 1000K tables scenario

[lidezhu]

What did you do?

Create 1000K tables and set up a changefeed to synchronize them.

What did you expect to see?

The changefeed can initialize normally, and the resolvedTs lag drops to the seconds level.

What did you see instead?

Changefeed initialization speed is really slow and maintainer handler latency is high.

Versions of the cluster

Upstream TiDB cluster version (execute SELECT tidb_version(); in a MySQL client):

(paste TiDB cluster version here)

Upstream TiKV version (execute tikv-server --version):

(paste TiKV version here)

TiCDC version (execute cdc version):

(paste TiCDC version here)