[csi] Add NodeStageVolume and NodeUnstageVolume

[AleksZimin]

Description

This PR introduces the NodeStageVolume and NodeUnstageVolume methods to our CSI driver. These methods enhance the handling of multi-attach scenarios for the same volume across multiple containers. Specifically, NodeStageVolume is responsible for preparing the volume, including filesystem creation (if needed) and performing the initial mount. Subsequently, the NodePublishVolume step performs a bind mount to make the volume available to the container. To ensure thread safety and prevent concurrent processing issues, we have also implemented a map that tracks volumes currently being processed.

Why do we need it, and what problem does it solve?

Adding NodeStageVolume and NodeUnstageVolume addresses the challenge of correctly managing volumes that are attached to multiple containers simultaneously, particularly in filesystem mode. Previously, handling simultaneous attachments could lead to errors or race conditions, especially when operations like filesystem creation were involved. By separating the mount and bind mount stages and introducing a tracking mechanism for volume processing, we enhance the robustness and reliability of our volume management.

What is the expected result?

• Enhanced support for multi-attach volume scenarios, ensuring that volumes attached to multiple containers are managed safely and efficiently.
• Prevention of race conditions and errors during volume setup, especially when performing filesystem operations.
• Improved overall reliability and performance of the CSI driver in complex deployment environments.

Checklist

• The code is covered by unit tests.
• e2e tests passed.
• Documentation updated according to the changes.
• Changes were tested in the Kubernetes cluster manually.