Add max_retries to RunnableNodeHandle.run and restart methods

Centralized retry logic was recently added to the core Node class to handle
transient startup errors automatically. This commit exposes that functionality
through the user-facing handles (ApiNode, InitNode, WitnessNode, etc.) by
adding the max_retries parameter to RunnableNodeHandle.

Additionally, added missing alternate_chain_specs parameter to RunnableNodeHandle.restart
to maintain parity with the underlying Node.restart method.

- Update RunnableNodeHandle.run to accept max_retries (default 1)
- Update RunnableNodeHandle.restart to accept alternate_chain_specs and max_retries
- Pass parameters down to the underlying Node implementation