feat: tcmalloc memory release improvements

[hycdong]

What is this pull request solved

In pull request #343 , we optimize tcmalloc release memory. Based on the former implementation, this pull request add more improvements about it:

1. Add a new perf-counter to record tcmalloc release memory
2. Add a new remote command to control release memory or not
3. Fix percentage range check from (0, 100) to (0, 100]

New perf-counter

replica*eon.replica_stub*tcmalloc.release.memory.size

This perf-counter has a problem that it will only be updated every interval time. For example, if tcmalloc memory release time interval is one hour, one server releases 1G memory at 10:00, the value won't change until 11:00. As a result, the monitor graph will be like a trapezoid.
If you have a better way to solve this problem, please comment this pull request or discuss with me offline.

New remote command

command help: replica.release-tcmalloc-memory <true|false>

Usage sample:

// query current value
>>> remote_command -t replica-server replica.release-tcmalloc-memory
COMMAND: replica.release-tcmalloc-memory

CALL [replica-server] [ip1: port1] succeed: true
CALL [replica-server] [ip2: port2] succeed: true
CALL [replica-server] [ip3: port3] succeed: true

Succeed count: 3
Failed count: 0

// not release memory
>>> remote_command -t replica-server replica.release-tcmalloc-memory false
COMMAND: replica.release-tcmalloc-memory false

CALL [replica-server] [ip1: port1] succeed: OK
CALL [replica-server] [ip2: port2] succeed: OK
CALL [replica-server] [ip3: port3] succeed: OK

Succeed count: 3
Failed count: 0

[neverchanje]

tcmalloc_release_memory_size