feat: Improve HTTP client caching, memory configuration and extension enhancements

[lcomplete]

Summary

This PR includes several improvements across the server and browser extension:

Server Improvements

• HTTP Client Caching: Enhanced HttpUtils with caching and client management for feed requests
• Memory Configuration: Increased JAVA_ARGS max heap size to 1024m in Dockerfile
• Connection Timeout: Ensured connection timeout is properly set in application.yml

Browser Extension Improvements

• Context Menu: Added DEV flag to context menu titles in development mode
• Article Preview: Simplified title rendering logic by removing isXTwitterSite function

Other Changes

• Added .agents directory to .gitignore

[augmentcode]

🤖 Augment PR Summary

Summary: This PR improves server-side feed fetching behavior and runtime configuration.

Changes:

• Reuses an `OkHttpClient` (keyed by proxy + timeout) and shares a single disk `Cache` for feed requests to reduce repeated client/cache construction.
• Normalizes RSS feed fetch/parse failure handling by wrapping `IOException`/`FeedException` in `ConnectorFetchException`.
• Closes Lucene `DirectoryReader` instances via try-with-resources in both indexing and searching paths.
• Adjusts runtime settings: bumps JVM max heap (`-Xmx`) to 1024m and sets Hikari `connection-timeout` to 60s.
• Adds `.agents/` to `.gitignore`.

Technical Notes: Feed client caching is now process-wide (static), so behavior depends on the proxy/timeout key and shared cache directory.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 2 suggestions posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

The proxy guard only checks host, but proxySetting.getPort() is unboxed when constructing InetSocketAddress and will NPE if the port is null. Consider also guarding port here to avoid crashing feed fetch when proxy config is partially populated.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

[augmentcode]

If a request passes page <= 0, then startIndex = (page - 1) * size and TopScoreDocCollector.create(page * size, ...) can go negative/zero and Lucene may throw at query time. Consider clamping page to at least 1 before using it in these calculations.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}