drainer: can't send request correctly to pump when set compressor=gzip

[lichunzhu]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. What did you do?
   If possible, provide a recipe for reproducing the error.
   Start a TiDB cluster and set compressor=grip in drainer's config.

2. What did you expect to see?
   Drainer can replicate data correctly.

3. What did you see instead?
   Drainer fails to pull binlogs from pump. Pump can't read request correctly:
   pump.log:

[2022/03/22 19:29:34.971 +08:00] [DEBUG] [server.go:271] ["get PullBinlogs request"] [request="{\"startFrom\":{}}"]
[2022/03/22 19:29:34.971 +08:00] [DEBUG] [server.go:273] ["PullBinlogs request quit"] [request="{\"startFrom\":{}}"]

4. Please provide the relate downstream type and version of drainer.
   (run drainer -V in terminal to get drainer's version)
   v5.2.3

[lichunzhu]

Guess it's because the pump doesn't set a decompressor.

[AmoebaProtozoa]

having similar issue on v5.2.2 after changing drainer's compressor option to "gzip"
compressor settings seems to be correctly applied but failed to find correct leader

[2022/03/29 18:08:22.919 +08:00] [INFO] [pump.go:138] ["pump create pull binlogs client"] [id=]
[2022/03/29 18:08:22.919 +08:00] [INFO] [pump.go:207] ["pump grpc compression enabled"] [id=]
[2022/03/29 18:08:22.920 +08:00] [ERROR] [pump.go:152] ["pump receive binlog failed"] [id=] [error="rpc error: code = Unknown desc = cluster ID are mismatch, 0 vs 7005077429661404245"]

[lichunzhu]

After a careful check this problem is introduced in #1039

In this PR we set pump/drainer's max grpc message size to max_int:

	maxGrpcMsgSize  = int(^uint(0) >> 1)

However, when grpc decompresses messages, it will left a few buffer for this, which overflows the int64:
https://github.com/grpc/grpc-go/blob/v1.44.0/rpc_util.go#L742