Add physical hardware information for admins

[achimnol]

In the admin GraphQL queries for agents, let's include physical hardware information including:

• CPU architecture, vendor, family, cores, frequency, feature flags
• Memory
• NUMA or not, number of NUMA nodes
• Accelerators (driver/runtime version, compute capability, processors and memory, etc.)
  • Include utilization reports like nvidia-smi
• Disk size

In conjunction with lablup/backend.ai-manager#103, let's add the followings to the agent:

• [x] Add collect_live_stats() abstract method to AbstractComputeDevice
• [ ] Add get_physical_info() abstract method to AbstractComputeDevice
• [x] Add collect_live_stats_summary() abstract method to AbstractComputePlugin
• [ ] Add get_physical_info_summary() abstract method to AbstractComputePlugin

All above methods should return an arbitrary JSON-serializable dict.

[achimnol]

Live stats and plugin-specific extra information is now implemented via lablup/backend.ai-agent#109 and related updates to the manager code. Let's elaborate hardware part a little bit more, to include specific model names for instance.

[achimnol]

The commit 0b135ec79 (#154) partially resolves this by storing additional "attached_device" information in the kernels table so that admins can inspect which type of GPU models are used for each kernel session.

Though, this issue is kept open since it targets providing the "current" information of physical devices of agents while the commit 0b135ec79 stores historical device usage information of kernel sessions.

[achimnol]

In the context of "H" project, this is now resolved.

[achimnol]

This is now implemented as gather_hwinfo() RPC API of agents.