child mount process should not exit when we kill the parent process.

[zhoucheng361]

What happened:
https://github.com/juicedata/juicefs/actions/runs/12496360301/job/34868010820

root@bench-01:~/juicefs# ps -ef | grep mount
tidb      375693       1  0 Dec18 ?        00:00:00 bin/node_exporter/node_exporter --web.listen-address=:9100 --collector.tcpstat --collector.mountstats --collector.meminfo_numa --collector.buddyinfo --collector.vmstat.fields=^.* --log.level=info
root     1130951       1  0 17:37 ?        00:00:00 ./juicefs mount sqlite3:///root/juicefs/test.db /tmp/jfs -d
root     1131018 1130951  1 17:37 ?        00:00:00 /root/juicefs/juicefs mount sqlite3:///root/juicefs/test.db /tmp/jfs -d
root     1131100 1130795  0 17:37 pts/4    00:00:00 grep --color=auto mount
root@bench-01:~/juicefs# kill 1130951
root@bench-01:~/juicefs# ps -ef | grep mount
tidb      375693       1  0 Dec18 ?        00:00:00 bin/node_exporter/node_exporter --web.listen-address=:9100 --collector.tcpstat --collector.mountstats --collector.meminfo_numa --collector.buddyinfo --collector.vmstat.fields=^.* --log.level=info
root     1131114 1130795  0 17:37 pts/4    00:00:00 grep --color=auto mount

What you expected to happen:
child process should not exit when we kill the parent process.

How to reproduce it (as minimally and precisely as possible):

Anything else we need to know?

Environment:

• JuiceFS version (use juicefs --version) or Hadoop Java SDK version:
• Cloud provider or hardware configuration running JuiceFS:
• OS (e.g cat /etc/os-release):
• Kernel (e.g. uname -a):
• Object storage (cloud provider and region, or self maintained):
• Metadata engine info (version, cloud provider managed or self maintained):
• Network connectivity (JuiceFS to metadata engine, JuiceFS to object storage):
• Others:

[zhoucheng361]

broken from this run: https://github.com/juicedata/juicefs/actions/runs/11468091920

[zhijian-pro]

In version 1.2, when the parent process receives a term signal, it will not pass it on to the child process, and the child process will become orphaned. After this PR, the parent process will pass the term signal to the child process, and the child process will exit proactively.

[davies]

We should revisit this before 1.3 release

[jiefenghuang]

The parent process manages the lifecycle of the service process, so sending signals here is a more reasonable approach. Does it need any further modifications?