Support setting health check interval for model service runtimes

[kyujin-cho]

Main idea

On current implementation we do not have a way to set interval of health check requests periodically sent by Kernel Runner (ai.backend.kernel). Let's extend the runner to support controlling the value.

• Add a new key definition on ai.backend.common.config.model_definition_iv which describes the interval of health check request
• Update kernel runner (and agent also) implementation to resepct the health check interval value set by user

Alternative ideas

No response

Anything else?

No response