Upgrade process use the wrong URL

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. What did you do?
   Upgrade cluster

2. What did you expect to see?
   Successfully

3. What did you see instead?

Upgrading component tikv
failed increasing schedule limit: no endpoint available, the last err was: error requesting http://10.4.101.105:2379/pd/api/v1/config/replication-mode, response: {
"code": "input",
"msg": "unexpected end of JSON input",
"data":

{ "Offset": 0 }

See https://github.com/pingcap/tiup/blob/49ee71a1e3efc77793fed1ddcaa37a6965ac5845/pkg/cluster/operation/upgrade.go#L223

https://github.com/pingcap/tiup/blob/49ee71a1e3efc77793fed1ddcaa37a6965ac5845/pkg/cluster/api/pdapi.go#L771

For xxx-schedule-limit, we should use pd/api/v1/config or pd/api/v1/config/schedule instead of pd/api/v1/config/replication-mode. replication-mode hasn't GA for now.

4. What version of TiUP are you using (tiup --version)?
   master

Start by reading pkg/cluster/operation/upgrade.go around line 223 and pkg/cluster/api/pdapi.go around line 771. Trace how the xxx-schedule-limit request is assembled, then verify that cluster upgrade uses the supported PD configuration endpoint and no longer produces the reported replication-mode error.