PD transfers leader to a pending peer

Please answer these questions before submitting your issue. Thanks!

1. What did you do?
```
./pd-ctl -u 10.233.100.247:2379 region 56317
{
"id": 56317,
"start_key": "7480000000000000FFF55F698000000000FF0000010130303030FF30303030FF303030FF3030303030FF3030FF303030303030FF30FF30303030303030FFFF3030303030303032FFFF34303231313336FF38FF000000000000FF0000F70400000000FF0E5155A801376162FF3366363833FF6639FF643766333135FF38FF63363065663435FFFF3139313933396236FFFF00000000000000FF00F7040000000000FF0000000400000000FF0000000000000000FB",
"end_key": "7480000000000000FFF55F698000000000FF0000010130303030FF30303030FF303030FF3030303030FF3030FF303030303030FF30FF30303030303030FFFF3030303030303032FFFF35303038343932FF34FF000000000000FF0000F70400000000FF0EE7FE3F01383431FF6330386162FF6530FF313031376536FF38FF39326137636363FFFF3033656138373462FFFF00000000000000FF00F7040000000000FF0000000400000000FF0000000000000000FB",
"epoch": {
"conf_ver": 37,
"version": 237
},
"peers": [
{
"id": 56318,
"store_id": 1
},
{
"id": 56319,
"store_id": 6
},
{
"id": 56320,
"store_id": 9
},
{
"id": 57144,
"store_id": 5
}
],
"leader": {
"id": 56318,
"store_id": 1
},
"pending_peers": [
{
"id": 56319,
"store_id": 6
}
],
"written_bytes": 19350993,
"approximate_size": 123,
"approximate_keys": 624077
}
```

2. What did you expect to see?

PD will do scheduling quickly.

3. What did you see instead?

PD tries to transfer leader for region 56317 from store 1 to store 6 which contains a pending peer, and it makes PD to repeatedly do this operation, which blocks the scheduling procedure.
[hot-region.log](https://github.com/pingcap/pd/files/3425649/hot-region.log)

4. What version of PD are you using (`pd-server -V`)?
```
Release Version: v3.0.0-2-gee98bf9
Git Commit Hash: ee98bf9063e9aef9e23e02fdf0a3b20b96e32d20
Git Branch: pingcap/release-3.0
UTC Build Time: 2019-07-11 03:50:52
```