fix: memory-leak-like issue in nv_trust attestation

[jcabrero]

This PR changes the code to make sure the nv_trust only creates the client once and adds the verifier once. Otherwise, we cause the attester to be attached to a global instance and to be recreated multiple times. This makes the attestation to take exponentially more time, as more verifiers produce the attestation (though one is only produced).

The code is reorganized to ensure the client is only created once.

[Copilot]

Pull Request Overview

This PR fixes a memory-leak-like performance issue in the NVIDIA Trust attestation system by ensuring the attestation client and verifier are created only once instead of being recreated multiple times, which was causing exponential slowdown.

• Centralizes client creation in a new get_client() function with @lru_cache to ensure singleton behavior
• Removes duplicate client configuration code from attester and verifier modules
• Consolidates shared functionality and constants into the __init__.py module

Reviewed Changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
packages/nilai-common/src/nilai_common/logger.py	Removes unused default logger instance
nilai-attestation/src/nilai_attestation/attestation/nvtrust/__init__.py	Adds centralized client creation with caching and shared utilities
nilai-attestation/src/nilai_attestation/attestation/nvtrust/nv_attester.py	Refactors to use centralized client creation and removes duplicate code
nilai-attestation/src/nilai_attestation/attestation/nvtrust/nv_verifier.py	Updates to use centralized client creation and removes hardcoded name parameter
nilai-attestation/src/nilai_attestation/app.py	Adds logging configuration to suppress verbose SDK output