feat(health): report redacted datasource/SSH integration errors

[PrashantBtkl]

Description

Adds error_class (datasource_integration_failure / ssh_integration_failure) to per-datasource entries in the health report, so the Proxy Agent tab can show why an integration is failing without logging into the host.

Errors come from failed health checks and failed proxied requests (kept up to 15 min, cleared by the next successful request). They are redacted (credentials, private keys, IPs, host:port) and truncated to 512 chars before leaving the host. Rather than scraping journald, the agent reports from its own in-process state, so it works on all platforms. Errors clear automatically once the integration recovers.

Relates to nudgebee/nudgebee-enterprise#39578 (backend/UI storage and rendering of error_class still to do).

Type of change

• New feature (non-breaking change which adds functionality)
• Documentation

How Has This Been Tested?

• Unit tests (go vet and go test ./pkg/...; make validate not run)

Checklist

• CLA signed (the CLA bot will prompt on your first PR)
• make validate passes (fmt + lint + test)
• Docs updated if the wire shape, config surface, or proxy module behavior changed

🤖 Generated with Claude Code

[gemini-code-assist]

Code Review

This pull request introduces integration error reporting for the proxy agent, adding error redaction, truncation, and tracking of recent request errors with a 15-minute TTL. Feedback on these changes highlights three key areas: first, string truncation in RedactError should safely handle multi-byte UTF-8 characters to prevent invalid sequences; second, a potential memory leak in Registry should be addressed by cleaning up reqErrors when datasources are removed or closed; and third, the test suite for RedactError should assert exact expected outputs to prevent tests from passing vacuously.