Add resilient_client for exponential backoff

[sarroutbi]

This pull request introduces a resilient HTTP client with an exponential backoff strategy to enhance the robustness of the Keylime push model agent. This new client is then integrated into the push-model agent's attestation process, making it more tolerant of transient network failures and temporary server-side errors.

1. Add Resilient HTTP Client with Exponential Backoff

The first commit adds a new, reusable module named resilient_client.

• New Module: Creates keylime/src/resilient_client.rs.
• Functionality: This module provides a ResilientClient that wraps reqwest. It uses the reqwest-retry and reqwest-middleware crates to automatically retry failed HTTP requests.
• Custom Retry Strategy: A custom retry strategy, StopOnSuccessStrategy, is implemented. This allows the client to intelligently stop retrying not only on failures but also when it receives a specific, configurable list of successful HTTP status codes (e.g., 200 OK, 202 Accepted).
• Configurability: The client is configurable with parameters for maximum retries, initial delay, and maximum delay, providing flexible control over the backoff behavior.
• Dependencies: Adds reqwest-middleware, reqwest-retry, and retry-policies to the project. All these newly introduced crates (and used versions) exist on Fedora:
  https://src.fedoraproject.org/rpms/rust-retry-policies
  https://src.fedoraproject.org/rpms/rust-reqwest-middleware
  https://src.fedoraproject.org/rpms/rust-reqwest-retry

2. Integrate Resilient Client into Push Attestation

The second commit applies the new ResilientClient to the agent's attestation workflow.

• Integration: The keylime-push-model-agent's AttestationClient is refactored to use the new ResilientClient instead of a standard reqwest::Client.
• Increased Robustness: HTTP calls for attestation negotiation and evidence submission will now automatically retry on connection errors or server-side statuses like 503 Service Unavailable.
• New Configuration: New configuration options have been added to the agent's config file (keylime.conf) to control the retry behavior:
  • expbackoff_max_retries
  • expbackoff_initial_delay
  • expbackoff_max_delay
• Improved Reliability: This change makes the push-model agent significantly more reliable in environments where the verifier or registrar might be temporarily unavailable.

[codecov]

Codecov Report

Attention: Patch coverage is 63.15789% with 28 lines in your changes missing coverage. Please review.

Project coverage is 58.86%. Comparing base (7b746b0) to head (933cd4d).
Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
keylime/src/resilient_client.rs	47.82%	24 Missing ⚠️
keylime-push-model-agent/src/attestation.rs	81.81%	4 Missing ⚠️

Additional details and impacted files

Flag	Coverage Δ
e2e-testsuite	58.86% <63.15%> (+0.31%)	⬆️
upstream-unit-tests	58.86% <63.15%> (+0.31%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
keylime-push-model-agent/src/main.rs	45.79% <100.00%> (+0.60%)	⬆️
keylime/src/config/base.rs	86.90% <100.00%> (+0.18%)	⬆️
keylime/src/config/push_model.rs	60.00% <ø> (ø)
keylime-push-model-agent/src/attestation.rs	39.69% <81.81%> (+17.55%)	⬆️
keylime/src/resilient_client.rs	47.82% <47.82%> (ø)

... and 7 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sarroutbi]

Test coverage continues being imprecise:

[ansasaki]

LGTM! I just put a single note.