feat: introduced multi_bulk_walk_cmd

[gmaiyani-crest]

Add multi_bulk_walk_cmd for Multi-Varbind SNMP Walking

Summary

Introduces multi_bulk_walk_cmd to walk multiple OID trees simultaneously with lexicographic mode control per varbind.

Why?

• bulk_walk_cmd supports only a single varbind at a time and already includes lexicographic mode handling.
  However, to achieve the same behavior for multiple varbinds (from different OID subtrees), each varbind must be walked separately in sequence.
• This leads to multiple SNMP calls and increased total runtime, especially when a maxRows limit or call count constraint is applied on the system.
  The new multi_bulk_walk_cmd addresses this by allowing multiple varbinds to be walked in parallel within a single session, maintaining lexicographic consistency while significantly reducing network roundtrips and improving overall efficiency.

Key Changes

• Added multi_bulk_walk_cmd function for parallel walking of multiple OID subtrees
• Implemented lexicographic mode control for each varbind
• Restored etingof's bulkCmd multi-varbind with lexicographic capability with modern async/await patterns
• Maintained full backward compatibility with single-varbind bulk_walk_cmd usage
• Implemented precise maxRows handling that counts actual OIDs (not iterations) with automatic truncation
• Added per-varbind completion tracking for independent state management

Performance

• With comparing Etingof’s bulkCmd (using varbinds=[ObjectType(ObjectIdentity("1.3.6")), ObjectType(ObjectIdentity("1.3.6"))], lexicographicMode=False, max_repetitions=70), we observed a total of 41,710 OIDs retrieved over 20,855 SNMP calls.
• Using the same configuration and restoring identical lexicographic behavior with the new multi_bulk_walk_cmd, the results were 41,708 OIDs retrieved with only 791 SNMP calls.
• The overall execution time improved by approximately 5–7 seconds faster under the same test conditions.
• When attempting to parallelize bulk_walk_cmd externally (i.e., running processes per OID subtree to simulate concurrency), The overall call count and runtime both increased when max_repetitions was large.
• for example, when multiple OIDs had only a few sub-OIDs, or when max_repetitions was comparable to the total subtree size. In such cases, the total execution time was observed to be 60–70% higher compared to multi_bulk_walk_cmd. When running on limited resources, this effect becomes more noticeable, adding extra latency and SNMP calls.

[CLAassistant]

All committers have signed the CLA.