fix: Resolve host.docker.internal hostname resolution issue

- Change cache host from 'host.docker.internal' to empty string
- Allow act_runner to auto-detect the correct host IP address
- Update all runner configs: docker, heavy, light, security
- Improve troubleshooting scripts with host IP detection:
  - Linux/macOS: Use ip route, hostname -I, or ifconfig
  - Windows: Use Get-NetIPAddress PowerShell cmdlets
- Update documentation to reflect auto-detection approach

This resolves the 'getaddrinfo ENOTFOUND host.docker.internal' error
by using a more compatible approach that works across different
Docker setups and operating systems.

runners/config_heavy.yaml

@@ -54,9 +54,9 @@ cache:
   # If it's empty, the cache data will be stored in $HOME/.cache/actcache.
   dir: ""
   # The host of the cache server.
-  # Use host.docker.internal to allow containers to access the host
-  # This fixes the common networking issue where containers can't reach the cache server
-  host: "host.docker.internal"
+  # Leave empty to auto-detect the host IP address
+  # This is more compatible across different Docker setups
+  host: ""
   # Use a fixed port instead of random to avoid connection issues
   port: 44029
   # The external cache server URL. Valid only when enable is true.