Sink operations can't be canceled correctly if network becomes unavailable

[hicqu]

What did you do?

1. Create a cdc capture with a MySQL downstream;
2. Drop all network packages to the downstream;
3. Run a workload on the upstream cluster.

What did you expect to see?

CDC can detect that network between the downstream and itself becomes broken. And then try to re-establish connections.

What did you see instead?

CDC can't detect the network problem. All sink workers are blocked.

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)