Reduce cgroups read frequency to avoid kernel kernfs clock pressure

[xigang]

Fixes ##3528

When GetInfo is called with false parameter, it now reads cgroups data once every 5 seconds and gets data from containerData cache at other times. This change reduces the frequency of reading cgroups to prevent potential Linux
machine hangs caused by kernel kernfs clock pressure.

The change affects the containerDataToContainerInfo method in manager.go which is used to retrieve container information. By caching and reusing the cgroups data between reads, we reduce system load while still maintaining
reasonably up-to-date container metrics.

[xigang]

/cc @dims @iwankgb @cwangVT PTAL.

[dims]

/ok-to-test

[xigang]

@dims Please run the workflow pipeline again, thanks.

# go test -v -run TestGetContainerInfo
=== RUN   TestGetContainerInfo
--- PASS: TestGetContainerInfo (0.00s)
=== RUN   TestGetContainerInfoV2
--- PASS: TestGetContainerInfoV2 (0.00s)
=== RUN   TestGetContainerInfoV2Failure
--- PASS: TestGetContainerInfoV2Failure (0.00s)
PASS
ok  	github.com/google/cadvisor/manager	0.010s

[xigang]

/retest

[xigang]

/cc @iwankgb Is there an issue with this workflow? It runs fine on my local setup.